index.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "926a030e",
   "metadata": {},
   "source": [
    "# **Mental Health Prediction for Kenyan university students**\n",
    "\n",
    "##### **Authors**:\n",
    "<ul type='square'> \n",
    "    <li> Alice Wamuyu</li>\n",
    "    <li> Eugene Kuloba </li>\n",
    "    <li> Fridah Kimathi </li>\n",
    "    <li> Karen Amanya  </li>\n",
    "    <li> Nicholus Magak  </li>\n",
    "    <li> Nobert Akwir </li>\n",
    "</ul>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c2668ecc",
   "metadata": {},
   "source": [
    "#  **1. Business Understanding**"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "24f72491",
   "metadata": {},
   "source": [
    "<img src=\"Images/mental-health.jpg\" alt=\"My Image\" width=\"700\" height=\"400\">"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "b1af7d93",
   "metadata": {},
   "source": [
    "#### **What is mental health ?**\n",
    "\n",
    "Mental health encompasses emotional, physiological and social well being. It affects how we feel, think, interact with others, handle stress and how we make choices.Good mental health is when a person is aware and utilizes his/her ability to deal with normal day to day life challenges whether at school , work and is productive in what they find fulfilling in life.Bad mental health is when a person finds it difficult to cope with how they are feeling, thinking or reacting to things and this can feel just as bad or even worse than physical illness.\n",
    "Mental health problems range from common problems such as anxiety and depression to rarer problems such as schizophrenia and bipolar disorder.Many factors contribute to mental health problems, including:\n",
    "\n",
    "  1. Biological factors, such as genes or brain chemistry\n",
    "  2. Life experiences, such as trauma or abuse\n",
    "  3. Family history of mental health problems\n",
    "\n",
    "#### **Why is mental health important**\n",
    "Mental health influences a variety of outcomes for an individual all the way to a nation. Healthier lifestyles, better physical health, fewer limitations in daily living, higher education, greater productivity in academics, employment and earnings, better relationships with others their overall quality of life are all influenced by positive mental health. Mental health problems are a common human experience and it's likely that,with a good combination of self-care, treatment and support that works for an individual they will get better.\n",
    "\n",
    "\n",
    "## **Problem Statement**\n",
    "<p>Despite resources in Kenyan universities growing immensely over the years, development of support services to students has not.Research has shown high levels of mental health problems among university students specifically depression and anxiety with the most affected group being students from poor backgrounds. The lack of or little provision for support services for such students results in dropouts and their inability to reach their full potential. </p>\n",
    "<p>The Africa Mental Health Foundation is an NGO that aims at conducting mental health research to get findings that aid in developing innovative practices for providing mental health services in Africa. In 2023, they are partnering with the ministry of Education in Kenya to develop a chatbot that will be available to all university students in Kenya. The goal of this is to  reduce the stigma around seeking help by providing the students with a platform where they can openly input their feelings / thoughts and this will attempt to classify the particular problem they are facing and therefore provide fitting resources.</p>\n",
    "<p>The use of technologies including machine learning and AI will potentially transform the delivery of mental health services in the coming years.This project aims to develop a machine learning model that classifies statements and questions expressed by university students in Kenya when speaking about the mental health challenges they struggle with and come up with a chatbot that will be used for a prototype of a mental health chatbot designed specifically for university students. </p>"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "c92dfede",
   "metadata": {},
   "source": [
    "## **Objectives**\n",
    "\n",
    "> ### **General Objective**\n",
    "The general objective is to leverage technology to make mental health services more accessible and more user-friendly for young people in Kenya and around the world.\n",
    "\n",
    "> ### **Specific Objectives**\n",
    "<ul type='square'>\n",
    "    <li > To develop a machine learning model that classifies statements and questions expressed by university students in Kenya when speaking about the mental health challenges they struggle with </li>\n",
    "    <li> To help universities establish mental health support and wellness services to their students. </li>\n",
    "    <li> To help university students in Kenya that are facing mental health problems to find resources and support services  that will enable them to get better  </li>\n",
    "</ul>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5adedfe",
   "metadata": {},
   "source": [
    "### **Importing the required libraries**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9650ea50",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "%config inlineBackend.figure_format = 'retina'\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import seaborn as sns\n",
    "sns.set_context('notebook')\n",
    "import string\n",
    "from textblob import TextBlob, Word\n",
    "import nltk\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.stem.porter import PorterStemmer\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier, AdaBoostClassifier\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.metrics import  accuracy_score, classification_report, log_loss"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "00eb33ab",
   "metadata": {},
   "source": [
    "# **2. Data Understanding**\n",
    "\n",
    "The data used in this project is from the <a href=\"https://zindi.africa/competitions/basic-needs-basic-rights-kenya-tech4mentalhealth/data\">  Basic Needs Basic Rights Kenya - Tech4MentalHealth</a> competition hosted by Zindi Africa. The data consists of statements and questions expressed by students from multiple universities across Kenya who reported suffering from these different mental health challenges. The wording of the statements is intended to respond to the prompting question, “What is on your mind?”"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f901a07b",
   "metadata": {},
   "source": [
    "#### **Loading the data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "df61ebb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = pd.read_csv('Data/Train.csv')\n",
    "validation_df = pd.read_csv('Data/Test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "03dde7a8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The train data shape: (616, 3)\n",
      "The test data shape: (309, 2)\n"
     ]
    }
   ],
   "source": [
    "# shape of the datasets\n",
    "print(f'The train data shape: {train_df.shape}')\n",
    "print(f'The test data shape: {validation_df.shape}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d126de24",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The train data columns: \n",
      " Index(['ID', 'text', 'label'], dtype='object') \n",
      "\n",
      "The test data columns: \n",
      " Index(['ID', 'text'], dtype='object')\n"
     ]
    }
   ],
   "source": [
    "# the columns in the datasets\n",
    "print(f'The train data columns: \\n {train_df.columns} \\n')\n",
    "print(f'The test data columns: \\n {validation_df.columns}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "df512500",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 616 entries, 0 to 615\n",
      "Data columns (total 3 columns):\n",
      " #   Column  Non-Null Count  Dtype \n",
      "---  ------  --------------  ----- \n",
      " 0   ID      616 non-null    object\n",
      " 1   text    616 non-null    object\n",
      " 2   label   616 non-null    object\n",
      "dtypes: object(3)\n",
      "memory usage: 14.6+ KB\n",
      "The train data info: None \n",
      " \n",
      "\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 309 entries, 0 to 308\n",
      "Data columns (total 2 columns):\n",
      " #   Column  Non-Null Count  Dtype \n",
      "---  ------  --------------  ----- \n",
      " 0   ID      309 non-null    object\n",
      " 1   text    309 non-null    object\n",
      "dtypes: object(2)\n",
      "memory usage: 5.0+ KB\n",
      "The test data info: None\n"
     ]
    }
   ],
   "source": [
    "# Checking for the shape, column names and data types \n",
    "print(f'The train data info: {train_df.info()} \\n \\n')\n",
    "print(f'The test data info: {validation_df.info()}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "a537cf2c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Depression    0.571429\n",
       "Alcohol       0.227273\n",
       "Suicide       0.107143\n",
       "Drugs         0.094156\n",
       "Name: label, dtype: float64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# classes proportionality \n",
    "train_df['label'].value_counts(normalize=True)\n",
    "\n",
    "    # There is a slight class imbalance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e47fa83a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualizing the proportionality of the target\n",
    "plt.title('Distribution of the Target Variable', fontsize = 18)\n",
    "plt.xlabel('Mental Health Problem')\n",
    "plt.ylabel('Count')\n",
    "train_df['label'].hist(figsize=(15,8));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "010a00bc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      39\n",
       "1      28\n",
       "2      57\n",
       "3      22\n",
       "4      51\n",
       "       ..\n",
       "611    36\n",
       "612    30\n",
       "613    24\n",
       "614    16\n",
       "615    31\n",
       "Name: length, Length: 616, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Creating a new column length showing the length of test per column\n",
    "train_df['length'] =  train_df['text'].apply(len)\n",
    "train_df['length']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b827835f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>616.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>39.813312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>21.438797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>8.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>26.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>35.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>48.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>196.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           length\n",
       "count  616.000000\n",
       "mean    39.813312\n",
       "std     21.438797\n",
       "min      8.000000\n",
       "25%     26.000000\n",
       "50%     35.000000\n",
       "75%     48.250000\n",
       "max    196.000000"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Statistical summary of the dataframe \n",
    "train_df.describe()\n",
    "\n",
    "# The smallest statement is 8 words long\n",
    "# The biggest statement is 196 words long"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "f8ab1159",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>194</th>\n",
       "      <td>J55053XP</td>\n",
       "      <td>I am financially constrained over school fees ...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>196</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           ID                                               text       label  \\\n",
       "194  J55053XP  I am financially constrained over school fees ...  Depression   \n",
       "\n",
       "     length  \n",
       "194     196  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Viewing the statement with the most words\n",
    "train_df[train_df['length'] == 196]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "444e546d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I am financially constrained over school fees and my  family background is not stable with a lot of debts…I have an elderly brother who could easily support me but has no job even after graduating\n"
     ]
    }
   ],
   "source": [
    "# Viewing the statement with the most words\n",
    "print(train_df['text'].iloc[194])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75832ca3",
   "metadata": {},
   "source": [
    " #### **Visualizing the distribution of the number of words per text**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "fa527957",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(15,8))\n",
    "sns.histplot(train_df['length'],ax=ax, kde=True)\n",
    "ax.set_ylabel(\"Count\")\n",
    "ax.set_xlabel(\"Number of words per text\")\n",
    "ax.set_title(\"The distribution of the number of words per text\", fontsize = 18);\n",
    "\n",
    "# Observations:\n",
    "    # The average text length lies between 25 and 40 words"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "02a16a1c",
   "metadata": {},
   "source": [
    " #### **Visualizing the effect of the number of words per text to the mental health status**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "8f9e2451",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "avg_text_length = train_df.groupby(\"label\").mean().sort_values(by='length', ascending=False)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(15,8))\n",
    "sns.barplot(x=avg_text_length.index,y=avg_text_length.length, ax=ax)\n",
    "ax.set_ylabel(\"Text Length\")\n",
    "ax.set_xlabel(\"Mental Health Status\")\n",
    "ax.set_title(\"The effect of the number of words per text to the mental health status\", fontsize = 18);\n",
    "\n",
    "# Observations:\n",
    "    #Depression has the highest number of words per text on average followed closely by Suicide.\n",
    "    # Alcohol and drugs have almost the same number of words per text which makes sense since alcohol is also a type of a drug."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ca10eea1",
   "metadata": {},
   "source": [
    "# **3. Data Preparation**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "212cd0b8",
   "metadata": {},
   "source": [
    " #### **i. Correcting spelling mistakes**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4ff34493",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "      <th>corrected_sent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SUAVK39Z</td>\n",
       "      <td>I feel that it was better I dieAm happy</td>\n",
       "      <td>Depression</td>\n",
       "      <td>39</td>\n",
       "      <td>I feel that it was better I die happy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9JDAGUV3</td>\n",
       "      <td>Why do I get hallucinations?</td>\n",
       "      <td>Drugs</td>\n",
       "      <td>28</td>\n",
       "      <td>Why do I get hallucinations?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>419WR1LQ</td>\n",
       "      <td>I am stresseed due to lack of financial suppor...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>57</td>\n",
       "      <td>I am stressed due to lack of financial support...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6UY7DX6Q</td>\n",
       "      <td>Why is life important?</td>\n",
       "      <td>Suicide</td>\n",
       "      <td>22</td>\n",
       "      <td>Why is life important?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FYC0FTFB</td>\n",
       "      <td>How could I be helped to go through the depres...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>51</td>\n",
       "      <td>Now could I be helped to go through the depres...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ID                                               text       label  \\\n",
       "0  SUAVK39Z            I feel that it was better I dieAm happy  Depression   \n",
       "1  9JDAGUV3                       Why do I get hallucinations?       Drugs   \n",
       "2  419WR1LQ  I am stresseed due to lack of financial suppor...  Depression   \n",
       "3  6UY7DX6Q                             Why is life important?     Suicide   \n",
       "4  FYC0FTFB  How could I be helped to go through the depres...  Depression   \n",
       "\n",
       "   length                                     corrected_sent  \n",
       "0      39              I feel that it was better I die happy  \n",
       "1      28                       Why do I get hallucinations?  \n",
       "2      57  I am stressed due to lack of financial support...  \n",
       "3      22                             Why is life important?  \n",
       "4      51  Now could I be helped to go through the depres...  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## To install textblob : conda install -c conda-forge textblob in terminal\n",
    "\n",
    "def correct_sent(text):\n",
    "    correction = TextBlob(text)\n",
    "    correction = correction.correct()\n",
    "    return str(correction)\n",
    "\n",
    "train_df['corrected_sent'] = train_df['text'].apply(lambda x: correct_sent(x))\n",
    "validation_df['corrected_sent'] = validation_df['text'].apply(lambda x: correct_sent(x))\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "769a2360",
   "metadata": {},
   "source": [
    " #### **ii. Changing text to lowercase**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a19c06b3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "      <th>corrected_sent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SUAVK39Z</td>\n",
       "      <td>I feel that it was better I dieAm happy</td>\n",
       "      <td>Depression</td>\n",
       "      <td>39</td>\n",
       "      <td>i feel that it was better i die happy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9JDAGUV3</td>\n",
       "      <td>Why do I get hallucinations?</td>\n",
       "      <td>Drugs</td>\n",
       "      <td>28</td>\n",
       "      <td>why do i get hallucinations?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>419WR1LQ</td>\n",
       "      <td>I am stresseed due to lack of financial suppor...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>57</td>\n",
       "      <td>i am stressed due to lack of financial support...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6UY7DX6Q</td>\n",
       "      <td>Why is life important?</td>\n",
       "      <td>Suicide</td>\n",
       "      <td>22</td>\n",
       "      <td>why is life important?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FYC0FTFB</td>\n",
       "      <td>How could I be helped to go through the depres...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>51</td>\n",
       "      <td>now could i be helped to go through the depres...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ID                                               text       label  \\\n",
       "0  SUAVK39Z            I feel that it was better I dieAm happy  Depression   \n",
       "1  9JDAGUV3                       Why do I get hallucinations?       Drugs   \n",
       "2  419WR1LQ  I am stresseed due to lack of financial suppor...  Depression   \n",
       "3  6UY7DX6Q                             Why is life important?     Suicide   \n",
       "4  FYC0FTFB  How could I be helped to go through the depres...  Depression   \n",
       "\n",
       "   length                                     corrected_sent  \n",
       "0      39              i feel that it was better i die happy  \n",
       "1      28                       why do i get hallucinations?  \n",
       "2      57  i am stressed due to lack of financial support...  \n",
       "3      22                             why is life important?  \n",
       "4      51  now could i be helped to go through the depres...  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df['corrected_sent'] = train_df['corrected_sent'].apply(lambda x: x.lower())\n",
    "validation_df['corrected_sent'] = validation_df['corrected_sent'].apply(lambda x: x.lower())\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7d199288",
   "metadata": {},
   "source": [
    " #### **iii. Removing the punctuation marks**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "37c1535f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "i feel hopeless, unworthy and useless …now do i cope with stress and forge the past?\n",
      "i am facing a lot of challenges in life financially, emotional, psycologically and with no solutions…now can i safely look for solutions about depression on goose\n",
      "there i get money for my needs…there do i get money  for personal needs?\n",
      "i am financially constrained over school fees and my  family background is not stable with a lot of debts…i have an elderly brother who could easily support me but has no job even after granulating\n",
      "i feel desperate…why is the world so unfair\n",
      "by relatives deny me…i wonder if i am part of my family?\n"
     ]
    }
   ],
   "source": [
    "# Checking texts with special characters such as â€¦ represented as …\n",
    "\n",
    "for x in train_df['corrected_sent']:\n",
    "    if '…' in x:\n",
    "        print(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "701a6dd2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Removing special characters â€¦ represented as …\n",
    "\n",
    "train_df['corrected_sent'] = train_df['corrected_sent'].apply(lambda x: x.replace('…', ' '))\n",
    "validation_df['corrected_sent'] = validation_df['corrected_sent'].apply(lambda x: x.replace('…', ' '))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "9cd439c0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "      <th>corrected_sent</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SUAVK39Z</td>\n",
       "      <td>I feel that it was better I dieAm happy</td>\n",
       "      <td>Depression</td>\n",
       "      <td>39</td>\n",
       "      <td>i feel that it was better i die happy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9JDAGUV3</td>\n",
       "      <td>Why do I get hallucinations?</td>\n",
       "      <td>Drugs</td>\n",
       "      <td>28</td>\n",
       "      <td>why do i get hallucinations</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>419WR1LQ</td>\n",
       "      <td>I am stresseed due to lack of financial suppor...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>57</td>\n",
       "      <td>i am stressed due to lack of financial support...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6UY7DX6Q</td>\n",
       "      <td>Why is life important?</td>\n",
       "      <td>Suicide</td>\n",
       "      <td>22</td>\n",
       "      <td>why is life important</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FYC0FTFB</td>\n",
       "      <td>How could I be helped to go through the depres...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>51</td>\n",
       "      <td>now could i be helped to go through the depres...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ID                                               text       label  \\\n",
       "0  SUAVK39Z            I feel that it was better I dieAm happy  Depression   \n",
       "1  9JDAGUV3                       Why do I get hallucinations?       Drugs   \n",
       "2  419WR1LQ  I am stresseed due to lack of financial suppor...  Depression   \n",
       "3  6UY7DX6Q                             Why is life important?     Suicide   \n",
       "4  FYC0FTFB  How could I be helped to go through the depres...  Depression   \n",
       "\n",
       "   length                                     corrected_sent  \n",
       "0      39              i feel that it was better i die happy  \n",
       "1      28                        why do i get hallucinations  \n",
       "2      57  i am stressed due to lack of financial support...  \n",
       "3      22                              why is life important  \n",
       "4      51  now could i be helped to go through the depres...  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# removing the other standard punctuation marks\n",
    "\n",
    "train_df['corrected_sent'] = train_df['corrected_sent'].apply(lambda x: x.strip(string.punctuation))\n",
    "validation_df['corrected_sent'] = validation_df['corrected_sent'].apply(lambda x: x.strip(string.punctuation))\n",
    "\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a3affb2",
   "metadata": {},
   "source": [
    " #### **iv. removing stop words**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "76e32181",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Downloading the necessary nltk packages. Uncomment to download\n",
    "\n",
    "#nltk.download('wordnet')\n",
    "#nltk.download('omw-1.4')\n",
    "#nltk.download('punkt')\n",
    "#nltk.download('stopwords')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "9d10351c",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "      <th>corrected_sent</th>\n",
       "      <th>no_stopwords</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SUAVK39Z</td>\n",
       "      <td>I feel that it was better I dieAm happy</td>\n",
       "      <td>Depression</td>\n",
       "      <td>39</td>\n",
       "      <td>i feel that it was better i die happy</td>\n",
       "      <td>feel better die happy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9JDAGUV3</td>\n",
       "      <td>Why do I get hallucinations?</td>\n",
       "      <td>Drugs</td>\n",
       "      <td>28</td>\n",
       "      <td>why do i get hallucinations</td>\n",
       "      <td>get hallucinations</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>419WR1LQ</td>\n",
       "      <td>I am stresseed due to lack of financial suppor...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>57</td>\n",
       "      <td>i am stressed due to lack of financial support...</td>\n",
       "      <td>stress due lack financial support school</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6UY7DX6Q</td>\n",
       "      <td>Why is life important?</td>\n",
       "      <td>Suicide</td>\n",
       "      <td>22</td>\n",
       "      <td>why is life important</td>\n",
       "      <td>life important</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FYC0FTFB</td>\n",
       "      <td>How could I be helped to go through the depres...</td>\n",
       "      <td>Depression</td>\n",
       "      <td>51</td>\n",
       "      <td>now could i be helped to go through the depres...</td>\n",
       "      <td>could help go depression</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ID                                               text       label  \\\n",
       "0  SUAVK39Z            I feel that it was better I dieAm happy  Depression   \n",
       "1  9JDAGUV3                       Why do I get hallucinations?       Drugs   \n",
       "2  419WR1LQ  I am stresseed due to lack of financial suppor...  Depression   \n",
       "3  6UY7DX6Q                             Why is life important?     Suicide   \n",
       "4  FYC0FTFB  How could I be helped to go through the depres...  Depression   \n",
       "\n",
       "   length                                     corrected_sent  \\\n",
       "0      39              i feel that it was better i die happy   \n",
       "1      28                        why do i get hallucinations   \n",
       "2      57  i am stressed due to lack of financial support...   \n",
       "3      22                              why is life important   \n",
       "4      51  now could i be helped to go through the depres...   \n",
       "\n",
       "                               no_stopwords  \n",
       "0                     feel better die happy  \n",
       "1                        get hallucinations  \n",
       "2  stress due lack financial support school  \n",
       "3                            life important  \n",
       "4                  could help go depression  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stopwords = nltk.corpus.stopwords.words('english')\n",
    "wordnet_lemmatizer = WordNetLemmatizer()\n",
    "\n",
    "def remove_stopwords(x):\n",
    "    sent = [wordnet_lemmatizer.lemmatize(i, 'v') for i in x.split() if i not in stopwords]\n",
    "    return ' '.join(sent)\n",
    "\n",
    "train_df['no_stopwords'] = train_df['corrected_sent'].apply(lambda x: remove_stopwords(x))\n",
    "validation_df['no_stopwords'] = validation_df['corrected_sent'].apply(lambda x: remove_stopwords(x))\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3d44fc8e",
   "metadata": {},
   "source": [
    "#### **iv. Performing train test split**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "dd8416e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      Depression\n",
       "1           Drugs\n",
       "2      Depression\n",
       "3         Suicide\n",
       "4      Depression\n",
       "          ...    \n",
       "611       Alcohol\n",
       "612       Suicide\n",
       "613       Alcohol\n",
       "614    Depression\n",
       "615    Depression\n",
       "Name: label, Length: 616, dtype: object"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_label = train_df['label']\n",
    "train_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "acd2dedc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      0\n",
       "1      3\n",
       "2      0\n",
       "3      2\n",
       "4      0\n",
       "      ..\n",
       "611    1\n",
       "612    2\n",
       "613    1\n",
       "614    0\n",
       "615    0\n",
       "Name: label, Length: 616, dtype: int64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Encoding the train data\n",
    "target={'Depression':0, 'Alcohol':1, 'Suicide':2, 'Drugs':3}\n",
    "\n",
    "train_label = train_label.map(target)\n",
    "train_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "1139b7b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(train_df, train_label, test_size=0.2 , random_state=42)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74338227",
   "metadata": {},
   "source": [
    " #### **v. Vectorization**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "f0b26b66",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>abandon</th>\n",
       "      <th>able</th>\n",
       "      <th>absent</th>\n",
       "      <th>abuse</th>\n",
       "      <th>academic</th>\n",
       "      <th>add</th>\n",
       "      <th>addict</th>\n",
       "      <th>addition</th>\n",
       "      <th>adduct</th>\n",
       "      <th>adults</th>\n",
       "      <th>...</th>\n",
       "      <th>world</th>\n",
       "      <th>worry</th>\n",
       "      <th>worst</th>\n",
       "      <th>worth</th>\n",
       "      <th>worthless</th>\n",
       "      <th>would</th>\n",
       "      <th>wrap</th>\n",
       "      <th>yet</th>\n",
       "      <th>young</th>\n",
       "      <th>youths</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 566 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   abandon  able  absent  abuse  academic  add  addict  addition  adduct  \\\n",
       "0        0     0       0      0         0    0       0         0       0   \n",
       "1        0     0       0      0         0    0       0         0       0   \n",
       "2        0     0       0      0         0    0       0         0       0   \n",
       "\n",
       "   adults  ...  world  worry  worst  worth  worthless  would  wrap  yet  \\\n",
       "0       0  ...      0      0      0      0          0      0     0    0   \n",
       "1       0  ...      0      0      0      0          0      0     0    0   \n",
       "2       0  ...      0      0      0      0          0      0     0    0   \n",
       "\n",
       "   young  youths  \n",
       "0      0       0  \n",
       "1      0       0  \n",
       "2      0       0  \n",
       "\n",
       "[3 rows x 566 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Instantiate countvectorizer\n",
    "countvec = CountVectorizer(ngram_range=(1,1))\n",
    "countvec1 = CountVectorizer(ngram_range=(1,1))\n",
    "\n",
    "# Vectorization of the data\n",
    "count_train_data = countvec.fit_transform(X_train.no_stopwords)\n",
    "count_test_data = countvec.transform(X_test.no_stopwords)\n",
    "\n",
    "count_training_data = countvec1.fit_transform(train_df.no_stopwords)\n",
    "count_validation_data = countvec1.transform(validation_df.no_stopwords)\n",
    "\n",
    "\n",
    "X_train= pd.DataFrame(count_train_data.toarray(), columns=countvec.get_feature_names())\n",
    "X_test  = pd.DataFrame(count_test_data.toarray(), columns=countvec.get_feature_names())\n",
    "\n",
    "train_data= pd.DataFrame(count_training_data.toarray(), columns=countvec1.get_feature_names())\n",
    "validation_data= pd.DataFrame(count_validation_data.toarray(), columns=countvec1.get_feature_names())\n",
    "\n",
    "X_train.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e19b53b5",
   "metadata": {},
   "source": [
    "# **3. Modelling**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "5991010a",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Models to be tested\n",
    "models = { 'Model' : ['Baseline Decision Tree', 'Baseline KNN Classifier', 'Baseline Random Forest Classifier',\\\n",
    "                      'Baseline Adaboost Classifier', 'Baseline Gradient Boost', 'baseline XGBoost Classifier',\\\n",
    "                        'XGBoost Classifier-Grid Search'],\n",
    "          'Train Accuracy Score(%)': [0, 0, 0, 0, 0, 0, 0],\n",
    "          'Test Accuracy Score(%)': [0, 0, 0, 0, 0, 0, 0],\n",
    "          'Train Log_loss': [0, 0, 0, 0, 0, 0, 0],\n",
    "          'Test Log_loss': [0, 0, 0, 0, 0, 0, 0] }\n",
    "          \n",
    "\n",
    "#Dataframe holding the model names and accuracy score and log_loss score\n",
    "df_model_results = pd.DataFrame(models, columns=['Model','Train Accuracy Score(%)', 'Test Accuracy Score(%)', 'Train Log_loss', 'Test Log_loss'])\n",
    "\n",
    "\n",
    "#Function to fill the dataframe holding model names and accuracy score\n",
    "\n",
    "def model_results(model_type, y_train, y_train_pred, y_train_pred_prob, y_test, y_test_pred, y_test_pred_prob):\n",
    "  index_val = df_model_results[df_model_results['Model']==model_type].index\n",
    "\n",
    "  df_model_results.loc[index_val, 'Train Accuracy Score(%)'] = round(accuracy_score(y_train, y_train_pred), 2)*100\n",
    "  df_model_results.loc[index_val, 'Test Accuracy Score(%)'] = round(accuracy_score(y_test, y_test_pred), 2)*100\n",
    "  df_model_results.loc[index_val, 'Train Log_loss'] = log_loss(y_train, y_train_pred_prob, labels=[0,1,2,3])\n",
    "  df_model_results.loc[index_val, 'Test Log_loss'] = log_loss(y_test, y_test_pred_prob, labels=[0,1,2,3])\n",
    "\n",
    "  return df_model_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "dde038ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Function to run all the baseline models:\n",
    "def baseline_models(model, model_name):\n",
    "    model.fit(X_train, y_train)\n",
    "\n",
    "    y_test_pred_prob = model.predict_proba(X_test)\n",
    "    y_train_pred_prob = model.predict_proba(X_train)\n",
    "\n",
    "    y_test_pred = model.predict(X_test)\n",
    "    y_train_pred = model.predict(X_train)\n",
    "\n",
    "    print('*********************************************************************')\n",
    "    print(classification_report(y_test, y_test_pred ))\n",
    "\n",
    "    print('*********************************************************************')\n",
    "    return model_results(model_name,y_train, y_train_pred, y_train_pred_prob, y_test, y_test_pred, y_test_pred_prob)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da93c6fa",
   "metadata": {},
   "source": [
    ">> <b>i. Decision Tree Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "eef8dfa6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.84      0.90      0.87        63\n",
      "           1       0.84      0.93      0.88        28\n",
      "           2       0.62      0.50      0.55        16\n",
      "           3       1.00      0.71      0.83        17\n",
      "\n",
      "    accuracy                           0.83       124\n",
      "   macro avg       0.82      0.76      0.78       124\n",
      "weighted avg       0.83      0.83      0.83       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.59081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                      0.0   \n",
       "2  Baseline Random Forest Classifier                      0.0   \n",
       "3       Baseline Adaboost Classifier                      0.0   \n",
       "4            Baseline Gradient Boost                      0.0   \n",
       "5        baseline XGBoost Classifier                      0.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334        5.59081  \n",
       "1                     0.0        0.000000        0.00000  \n",
       "2                     0.0        0.000000        0.00000  \n",
       "3                     0.0        0.000000        0.00000  \n",
       "4                     0.0        0.000000        0.00000  \n",
       "5                     0.0        0.000000        0.00000  \n",
       "6                     0.0        0.000000        0.00000  "
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline_dt = DecisionTreeClassifier(random_state=100)\n",
    "baseline_models(baseline_dt, 'Baseline Decision Tree')\n",
    "\n",
    "#Observation:\n",
    "     # Despite the fact that suicide has more data than drugs, it performs worse than it\n",
    "     # There is overfitting "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e616e870",
   "metadata": {},
   "source": [
    ">> <b>ii. KNN Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "46eca888",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.73      0.90      0.81        63\n",
      "           1       0.63      0.86      0.73        28\n",
      "           2       0.50      0.06      0.11        16\n",
      "           3       0.83      0.29      0.43        17\n",
      "\n",
      "    accuracy                           0.70       124\n",
      "   macro avg       0.67      0.53      0.52       124\n",
      "weighted avg       0.69      0.70      0.65       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                      0.0   \n",
       "3       Baseline Adaboost Classifier                      0.0   \n",
       "4            Baseline Gradient Boost                      0.0   \n",
       "5        baseline XGBoost Classifier                      0.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                     0.0        0.000000       0.000000  \n",
       "3                     0.0        0.000000       0.000000  \n",
       "4                     0.0        0.000000       0.000000  \n",
       "5                     0.0        0.000000       0.000000  \n",
       "6                     0.0        0.000000       0.000000  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline_knn_pipe = Pipeline(steps=[('sc', StandardScaler()), ('knn', KNeighborsClassifier())])\n",
    "baseline_models(baseline_knn_pipe, 'Baseline KNN Classifier')\n",
    "# Observation\n",
    " # KNN performs worse than decision tree"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cdec4664",
   "metadata": {},
   "source": [
    ">> <b>iii. Random Forest Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "f4d38f9c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.83      0.94      0.88        63\n",
      "           1       0.76      0.89      0.82        28\n",
      "           2       0.75      0.38      0.50        16\n",
      "           3       0.92      0.65      0.76        17\n",
      "\n",
      "    accuracy                           0.81       124\n",
      "   macro avg       0.81      0.71      0.74       124\n",
      "weighted avg       0.82      0.81      0.80       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>99.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0.106574</td>\n",
       "      <td>0.775972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                     99.0   \n",
       "3       Baseline Adaboost Classifier                      0.0   \n",
       "4            Baseline Gradient Boost                      0.0   \n",
       "5        baseline XGBoost Classifier                      0.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                    81.0        0.106574       0.775972  \n",
       "3                     0.0        0.000000       0.000000  \n",
       "4                     0.0        0.000000       0.000000  \n",
       "5                     0.0        0.000000       0.000000  \n",
       "6                     0.0        0.000000       0.000000  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline_rf = RandomForestClassifier(random_state=0)\n",
    "baseline_models(baseline_rf, 'Baseline Random Forest Classifier')\n",
    "\n",
    "# Observation \n",
    " # Random Forest performs better than Decision Tree and KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e73af31",
   "metadata": {},
   "source": [
    ">> <b>iv. Adaboost Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "cdd729a2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.74      0.95      0.83        63\n",
      "           1       0.92      0.79      0.85        28\n",
      "           2       0.67      0.25      0.36        16\n",
      "           3       0.69      0.53      0.60        17\n",
      "\n",
      "    accuracy                           0.77       124\n",
      "   macro avg       0.75      0.63      0.66       124\n",
      "weighted avg       0.76      0.77      0.74       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>99.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0.106574</td>\n",
       "      <td>0.775972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>85.0</td>\n",
       "      <td>77.0</td>\n",
       "      <td>1.031080</td>\n",
       "      <td>1.059045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                     99.0   \n",
       "3       Baseline Adaboost Classifier                     85.0   \n",
       "4            Baseline Gradient Boost                      0.0   \n",
       "5        baseline XGBoost Classifier                      0.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                    81.0        0.106574       0.775972  \n",
       "3                    77.0        1.031080       1.059045  \n",
       "4                     0.0        0.000000       0.000000  \n",
       "5                     0.0        0.000000       0.000000  \n",
       "6                     0.0        0.000000       0.000000  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline_adaboost = AdaBoostClassifier(random_state=0)\n",
    "baseline_models(baseline_adaboost, 'Baseline Adaboost Classifier')\n",
    "\n",
    "# Observations:\n",
    " # Adaboost performs worse than all the previous models\n",
    " # There is overfitting\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c66f469f",
   "metadata": {},
   "source": [
    ">> <b>v. Gradient Boost Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "bd52b9be",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.82      0.94      0.87        63\n",
      "           1       0.89      0.89      0.89        28\n",
      "           2       0.58      0.44      0.50        16\n",
      "           3       1.00      0.71      0.83        17\n",
      "\n",
      "    accuracy                           0.83       124\n",
      "   macro avg       0.82      0.74      0.77       124\n",
      "weighted avg       0.83      0.83      0.82       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>99.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0.106574</td>\n",
       "      <td>0.775972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>85.0</td>\n",
       "      <td>77.0</td>\n",
       "      <td>1.031080</td>\n",
       "      <td>1.059045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>97.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.163581</td>\n",
       "      <td>0.509805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                     99.0   \n",
       "3       Baseline Adaboost Classifier                     85.0   \n",
       "4            Baseline Gradient Boost                     97.0   \n",
       "5        baseline XGBoost Classifier                      0.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                    81.0        0.106574       0.775972  \n",
       "3                    77.0        1.031080       1.059045  \n",
       "4                    83.0        0.163581       0.509805  \n",
       "5                     0.0        0.000000       0.000000  \n",
       "6                     0.0        0.000000       0.000000  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline_gradientboost = GradientBoostingClassifier(random_state=0)\n",
    "baseline_models(baseline_gradientboost, 'Baseline Gradient Boost')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79d73e10",
   "metadata": {},
   "source": [
    ">> <b>vi. XGBoost Classifier</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "93a2a2d0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.83      0.94      0.88        63\n",
      "           1       0.89      0.89      0.89        28\n",
      "           2       0.71      0.62      0.67        16\n",
      "           3       1.00      0.65      0.79        17\n",
      "\n",
      "    accuracy                           0.85       124\n",
      "   macro avg       0.86      0.78      0.81       124\n",
      "weighted avg       0.85      0.85      0.84       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>99.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0.106574</td>\n",
       "      <td>0.775972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>85.0</td>\n",
       "      <td>77.0</td>\n",
       "      <td>1.031080</td>\n",
       "      <td>1.059045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>97.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.163581</td>\n",
       "      <td>0.509805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>93.0</td>\n",
       "      <td>85.0</td>\n",
       "      <td>0.219129</td>\n",
       "      <td>0.451154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                     99.0   \n",
       "3       Baseline Adaboost Classifier                     85.0   \n",
       "4            Baseline Gradient Boost                     97.0   \n",
       "5        baseline XGBoost Classifier                     93.0   \n",
       "6     XGBoost Classifier-Grid Search                      0.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                    81.0        0.106574       0.775972  \n",
       "3                    77.0        1.031080       1.059045  \n",
       "4                    83.0        0.163581       0.509805  \n",
       "5                    85.0        0.219129       0.451154  \n",
       "6                     0.0        0.000000       0.000000  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgboost_classifier = XGBClassifier()\n",
    "baseline_models(xgboost_classifier, 'baseline XGBoost Classifier')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3158b48",
   "metadata": {},
   "source": [
    ">> <b> Tuning XGboost model</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "a97e6844",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*********************************************************************\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.84      0.94      0.89        63\n",
      "           1       0.86      0.89      0.88        28\n",
      "           2       0.71      0.62      0.67        16\n",
      "           3       1.00      0.65      0.79        17\n",
      "\n",
      "    accuracy                           0.85       124\n",
      "   macro avg       0.85      0.78      0.80       124\n",
      "weighted avg       0.85      0.85      0.84       124\n",
      "\n",
      "*********************************************************************\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Train Accuracy Score(%)</th>\n",
       "      <th>Test Accuracy Score(%)</th>\n",
       "      <th>Train Log_loss</th>\n",
       "      <th>Test Log_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Baseline Decision Tree</td>\n",
       "      <td>99.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.012334</td>\n",
       "      <td>5.590810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Baseline KNN Classifier</td>\n",
       "      <td>82.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.512985</td>\n",
       "      <td>2.842548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Baseline Random Forest Classifier</td>\n",
       "      <td>99.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0.106574</td>\n",
       "      <td>0.775972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Baseline Adaboost Classifier</td>\n",
       "      <td>85.0</td>\n",
       "      <td>77.0</td>\n",
       "      <td>1.031080</td>\n",
       "      <td>1.059045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Baseline Gradient Boost</td>\n",
       "      <td>97.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>0.163581</td>\n",
       "      <td>0.509805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>baseline XGBoost Classifier</td>\n",
       "      <td>93.0</td>\n",
       "      <td>85.0</td>\n",
       "      <td>0.219129</td>\n",
       "      <td>0.451154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost Classifier-Grid Search</td>\n",
       "      <td>92.0</td>\n",
       "      <td>85.0</td>\n",
       "      <td>0.241682</td>\n",
       "      <td>0.439006</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               Model  Train Accuracy Score(%)  \\\n",
       "0             Baseline Decision Tree                     99.0   \n",
       "1            Baseline KNN Classifier                     82.0   \n",
       "2  Baseline Random Forest Classifier                     99.0   \n",
       "3       Baseline Adaboost Classifier                     85.0   \n",
       "4            Baseline Gradient Boost                     97.0   \n",
       "5        baseline XGBoost Classifier                     93.0   \n",
       "6     XGBoost Classifier-Grid Search                     92.0   \n",
       "\n",
       "   Test Accuracy Score(%)  Train Log_loss  Test Log_loss  \n",
       "0                    83.0        0.012334       5.590810  \n",
       "1                    70.0        0.512985       2.842548  \n",
       "2                    81.0        0.106574       0.775972  \n",
       "3                    77.0        1.031080       1.059045  \n",
       "4                    83.0        0.163581       0.509805  \n",
       "5                    85.0        0.219129       0.451154  \n",
       "6                    85.0        0.241682       0.439006  "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tuned_xgboost = XGBClassifier()\n",
    "param_grid = {'learning_rate': [0.1,0.2],\n",
    "               'max_depth': [4, 6, 10],\n",
    "               'n_estimators':[200, 300, 400]}\n",
    "\n",
    "\n",
    "xgbclassifier_grid_search = GridSearchCV(Tuned_xgboost, param_grid, scoring='accuracy', cv=3, n_jobs=1)\n",
    "\n",
    "xgbclassifier_grid_search.fit(X_train, y_train)\n",
    "\n",
    "y_test_pred_prob = xgbclassifier_grid_search.predict_proba(X_test)\n",
    "y_train_pred_prob = xgbclassifier_grid_search.predict_proba(X_train)\n",
    "\n",
    "y_test_pred = xgbclassifier_grid_search.predict(X_test)\n",
    "y_train_pred = xgbclassifier_grid_search.predict(X_train)\n",
    "\n",
    "print('*********************************************************************')\n",
    "print(classification_report(y_test, y_test_pred ))\n",
    "\n",
    "print('*********************************************************************')\n",
    "model_results('XGBoost Classifier-Grid Search',y_train, y_train_pred, y_train_pred_prob, y_test, y_test_pred, y_test_pred_prob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "c0c00471",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'learning_rate': 0.1, 'max_depth': 4, 'n_estimators': 300}"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# XGBoost Best Parameters\n",
    "xgbclassifier_grid_search.best_params_"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "059ca19b",
   "metadata": {},
   "source": [
    "# Final Model\n",
    "\n",
    "<i>The chosen  final model is the XGBoost Classifier</i>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "99b77b7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_model = XGBClassifier(learning_rate=0.1, max_depth=4, n_estimators=300)\n",
    " \n",
    "final_model.fit(train_data, train_label)\n",
    "\n",
    "preds = final_model.predict_proba(validation_data)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "88cf47e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "id = pd.DataFrame(validation_df['ID'])\n",
    "predicted = pd.DataFrame(preds, columns=['Depression', 'Alcohol', 'Suicide', 'Drugs'])\n",
    "predicted_status = pd.concat([id,predicted], axis=1)\n",
    "\n",
    "#Saving file as a CSV\n",
    "predicted_status.to_csv('Data/predicted_status.csv',encoding='utf-8',sep=',', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d62211aa",
   "metadata": {},
   "source": [
    "# **4. Evaluation**"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "b2aadf70",
   "metadata": {},
   "source": [
    "Based on the accuracy of our best model, the XGBoost Classifier-Grid Search, had a train accuracy score of 92% and test accuracy of 85%. The same model had a train and test log score of 0.242 and 0.439, respectively. We chose to use log loss as a performance metric because it takes into account the probabilities underlying in the model, not only the final output of the classification. This means, the higher the probabilities the lower the log loss value, which is better because it is basically a measure of uncertainty and the lower it is, the more certain we are of the potential occurrence. \n",
    "\n",
    "Since XGBoost classifier had the lowest log loss score, and it was chosen to be the best model, even though we had prior models with high training and test data accuracy.\n",
    "\n",
    "#### **_Findings_**:\n",
    "1. The dataset featured more observations of Depression and Anxiety than other classifications as a result the models used had lower recall and f1 scores in the classification of Drug and Alcohol related problems\n",
    "2. All models performed fairly well in classifying mental health problems in both the training and testing sets although accuracy was higher on the training data.\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "fdc9cb83",
   "metadata": {},
   "source": [
    "#### **_Limitations_**: \n",
    "i. Due to the imbalance in the dataset the F1-score for drug and alcohol was notably lower than depression and suicide.\n",
    "\n",
    "ii. Due to a relatively small dataset we could not achieve high score as desired."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "322b8d28",
   "metadata": {},
   "source": [
    "# **Conclusion and recommendations**"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "b6a7d9a8",
   "metadata": {},
   "source": [
    "The final model performed fairly well with a log score of 0.52 and an accuracy of around 85%. The Africa Mental Health Foundation(AMHF) can integrate the model into a chatbot prototype to carry out tests on actual university students and predict the student's mental state in order to assist them accordingly. This will help correctly identify and classify mental health problems to match students with the suitable resources. Moreover, the developed chatbot will encourage more students to  openly seek help without the fear of stigmatization."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "90744650",
   "metadata": {},
   "source": [
    "\n",
    "#### **_Recommendations_**\n",
    "From our model's ability to accurately identify and classify mental health problems at least 85% of the time. We would recommend that:\n",
    "\n",
    "1. The Africa Mental Health Foundation(AMHF) should collect more data that features drug and alcohol related problems as well as other mental health problems that have not been featured in the observations used here.\n",
    "\n",
    "2. The model be integrated into the chatbot prototype to carry out tests on actual university students and collect data on how it performed in classifying problems they were facing.\n",
    "\n",
    "3. The chatbot should also feature a database containing resources and services available to the students based on the problem that the model was able to identify. This will ensure that actual help is availed to the user."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "learn-env",
   "language": "python",
   "name": "learn-env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "vscode": {
   "interpreter": {
    "hash": "15e1ea0e9ae73a04f4db62dd1e96b3b16ca57ecdaaf21916cbbe9da1fa870ff9"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}