diff --git a/R/features-exp.R b/R/features-exp.R
index 44eaf38..88392d9 100644
--- a/R/features-exp.R
+++ b/R/features-exp.R
@@ -6,16 +6,19 @@ extract_features_exp <- function(x) {
   x <- x %>%
     dplyr::group_by(user_id) %>%
     dplyr::summarise(
+      ## tweets features
       n_sincelast = count_mean(since_last(.data$created_at)),
       n_timeofday = count_mean(hourofweekday(.data$created_at)),
       n = dplyr::n(),
       n_retweets = sum_(.data$is_retweet),
       n_quotes = sum_(.data$is_quote),
+      n_langs = tfse::n_uq(.data$lang),
       retweet_count = mean_(c(0, .data$retweet_count)),
       favorite_count = mean_(c(0, .data$favorite_count)),
-      favourites_count = max_(c(0, .data$favourites_count)),
       n_tweets = sum_(!.data$is_retweet & !.data$is_quote),
-
+      n_places = sum_(!is.na(.data$place_name)),
+      n_geo_coords = ncoord(.data$geo_coords),
+      n_bbox_coords = ncoord(.data$bbox_coords),
       iphone = sum_("Twitter for iPhone" %in% .data$source) / .data$n,
       webclient = sum_("Twitter Web Client" %in% .data$source) / .data$n,
       android = sum_("Twitter for Android" %in% .data$source) / .data$n,
@@ -25,28 +28,32 @@ extract_features_exp <- function(x) {
       google = sum_("Google" %in% .data$source) / .data$n,
       ifttt = sum_("IFTTT" %in% .data$source) / .data$n,
       facebook = sum_("Facebook" %in% .data$source) / .data$n,
-
       twittbotnet = sum_("twittbot.net" %in% .data$source) / .data$n,
       tweetdeck = sum_("TweetDeck" %in% .data$source) / .data$n,
-      twitterforblackberry = sum_("Twitter for BlackBerry®" %in% .data$source) / .data$n,
+      twitterforblackberry = sum_(
+        "Twitter for BlackBerry®" %in% .data$source) / .data$n,
       dlvrit = sum_("dlvr.it" %in% .data$source) / .data$n,
       instagram = sum_("Instagram" %in% .data$source) / .data$n,
       curiouscat = sum_("Curious Cat" %in% .data$source) / .data$n,
       echofon = sum_("Echofon" %in% .data$source) / .data$n,
-      ubersocialforblackberry = sum_("UberSocial for BlackBerry" %in% .data$source) / .data$n,
+      ubersocialforblackberry = sum_(
+        "UberSocial for BlackBerry" %in% .data$source) / .data$n,
       athkarapp = sum_("athkarApp" %in% .data$source) / .data$n,
       mobilewebm2 = sum_("Mobile Web (M2)" %in% .data$source) / .data$n,
       twitterfeed = sum_("twitterfeed" %in% .data$source) / .data$n,
       tweetbotforiοs = sum_("Tweetbot for iΟS" %in% .data$source) / .data$n,
-      tweetcasterforandroid = sum_("TweetCaster for Android" %in% .data$source) / .data$n,
-      twitcomcomunidades = sum_("Twitcom - Comunidades " %in% .data$source) / .data$n,
+      tweetcasterforandroid = sum_(
+        "TweetCaster for Android" %in% .data$source) / .data$n,
+      twitcomcomunidades = sum_(
+        "Twitcom - Comunidades " %in% .data$source) / .data$n,
       cloudhopper = sum_("Cloudhopper" %in% .data$source) / .data$n,
       twicca = sum_("twicca" %in% .data$source) / .data$n,
       wordpresscom = sum_("WordPress.com" %in% .data$source) / .data$n,
       mobileweb = sum_("Mobile Web" %in% .data$source) / .data$n,
       foursquare = sum_("Foursquare" %in% .data$source) / .data$n,
       showroomlive = sum_("SHOWROOM-LIVE" %in% .data$source) / .data$n,
-      twitterforwebsites = sum_("Twitter for Websites" %in% .data$source) / .data$n,
+      twitterforwebsites = sum_(
+        "Twitter for Websites" %in% .data$source) / .data$n,
       ios = sum_("iOS" %in% .data$source) / .data$n,
       tumblr = sum_("Tumblr" %in% .data$source) / .data$n,
       tweetlogix = sum_("Tweetlogix" %in% .data$source) / .data$n,
@@ -54,8 +61,10 @@ extract_features_exp <- function(x) {
       buffer = sum_("Buffer" %in% .data$source) / .data$n,
       twitcleplus = sum_("twitcle plus" %in% .data$source) / .data$n,
       keitaiweb = sum_("Keitai Web" %in% .data$source) / .data$n,
-      sandaysoftcumulus = sum_("Sandaysoft Cumulus" %in% .data$source) / .data$n,
-      twitpaneforandroid = sum_("TwitPane for Android" %in% .data$source) / .data$n,
+      sandaysoftcumulus = sum_(
+        "Sandaysoft Cumulus" %in% .data$source) / .data$n,
+      twitpaneforandroid = sum_(
+        "TwitPane for Android" %in% .data$source) / .data$n,
       playstationr4 = sum_("PlayStation(R)4" %in% .data$source) / .data$n,
       writelonger = sum_("Write Longer" %in% .data$source) / .data$n,
       featherforios = sum_("feather for iOS  " %in% .data$source) / .data$n,
@@ -66,28 +75,50 @@ extract_features_exp <- function(x) {
       janetter = sum_("Janetter" %in% .data$source) / .data$n,
       dynamictweets = sum_("Dynamic Tweets" %in% .data$source) / .data$n,
       twitcasting = sum_("TwitCasting" %in% .data$source) / .data$n,
-      ubersocialforandroid = sum_("UberSocial for Android" %in% .data$source) / .data$n,
-      janetterforandroid = sum_("Janetter for Android" %in% .data$source) / .data$n,
-      twitterforandroidtablets = sum_("Twitter for Android Tablets" %in% .data$source) / .data$n,
+      ubersocialforandroid = sum_(
+        "UberSocial for Android" %in% .data$source) / .data$n,
+      janetterforandroid = sum_(
+        "Janetter for Android" %in% .data$source) / .data$n,
+      twitterforandroidtablets = sum_(
+        "Twitter for Android Tablets" %in% .data$source) / .data$n,
       twitterformac = sum_("Twitter for Mac" %in% .data$source) / .data$n,
 
+      ## users features
+      lang_und = as.integer(.data$account_lang[1] == "und"),
+      lang_tr = as.integer(.data$account_lang[1] == "tr"),
+      lang_ru = as.integer(.data$account_lang[1] == "ru"),
+      lang_pt = as.integer(.data$account_lang[1] == "pt"),
+      lang_ja = as.integer(.data$account_lang[1] == "ja"),
+      lang_in = as.integer(.data$account_lang[1] == "in"),
+      lang_fr = as.integer(.data$account_lang[1] == "fr"),
+      lang_es = as.integer(.data$account_lang[1] == "es"),
+      lang_en = as.integer(.data$account_lang[1] == "en"),
+      lang_are = as.integer(.data$account_lang[1] == "ar"),
+      lang_de = as.integer(.data$account_lang[1] == "de"),
+      lang_it = as.integer(.data$account_lang[1] == "it"),
+      lang_id = as.integer(.data$account_lang[1] == "id"),
+      lang_ko = as.integer(.data$account_lang[1] == "ko"),
+      lang_nl = as.integer(.data$account_lang[1] == "nl"),
+      lang_hi = as.integer(.data$account_lang[1] == "hi"),
+      lang_fil = as.integer(.data$account_lang[1] == "fil"),
+      lang_th = as.integer(.data$account_lang[1] == "th"),
+      lang_engb = as.integer(.data$account_lang[1] == "en-gb"),
+      screen_name_alpha = nchar_(.data$screen_name[1]),
+      screen_name_num = ndigit_(.data$screen_name[1]),
       prof_image_na = sum_(is.na(.data$profile_image_url[1])),
       prof_image_type = sum_(grepl("\\.jpg", .data$profile_image_url[1])),
-
       profile_bg_na = sum_(is.na(.data$profile_background_url[1])),
       profile_bg_type = sum_(grepl("\\.png", .data$profile_background_url[1])),
-
       profile_bn_na = sum_(is.na(.data$profile_banner_url[1])),
-
       verified = as.integer(.data$verified[1]),
+      profile_url = !is.na(.data$profile_url[1]),
       years_on_twitter = relative_twitter_age(.data$account_created_at[1]),
       tweets_per_year = .data$n_tweets / (1 + .data$years_on_twitter),
-
-      ## i added one here so it wouldn't return NaN or undefined values (0 / x)
       statuses_count = max_(c(0, .data$statuses_count)),
       followers_count = max_(c(0, .data$followers_count)),
       friends_count = max_(c(0, .data$friends_count)),
       listed_count = max_(c(0, .data$listed_count)),
+      favourites_count = max_(c(0, .data$favourites_count)),
       tweets_to_followers = (.data$statuses_count + 1) /
         (.data$followers_count + 1),
       statuses_rate = (.data$statuses_count + 1) /
@@ -106,8 +137,15 @@ age_of_twitter <- function() {
 }
 
 relative_twitter_age <- function(account_created_at) {
-  years <- as.numeric(difftime(Sys.time(), account_created_at, units = "days"))/365
+  years <- as.numeric(difftime(
+    Sys.time(), account_created_at, units = "days"))/365
   aot <- age_of_twitter()
   ## set it at 15
   (years / aot) * 15
 }
+
+ncoord <- function(x) {
+  sum(vapply(x, function(.x) !is.na(.x[1]), integer(1), USE.NAMES = FALSE))
+}
+
+
diff --git a/R/utils.R b/R/utils.R
index 5a2f027..f6a16cb 100644
--- a/R/utils.R
+++ b/R/utils.R
@@ -67,3 +67,14 @@ count_mean <- function(x) {
   x <- as.integer(x) - 1L
   mean(x, na.rm = TRUE)
 }
+
+
+nchar_ <- function(x) {
+  ifelse(is.na(x), 0, nchar(x))
+}
+
+
+ndigit_ <- function(x) {
+  ifelse(is.na(x), 0, nchar(gsub("\\D", "", x)))
+}
+