What is a model explainer? A model explainer is a function designed to show how the model produces different predictions. Models in 2D are relatively easy to describe. For example, there we could have a classification model determined by a single line.
However, models frequently have more than 2 features, and so cannot be easily placed on a single chart. However, a model explainer can help us answer the following questions. Which features are the most important? How does changing those features impact the prediction? Is the model more complicated than it needs to be?
This information allows us to see inside the Blackbox of what a ML model is doing.
https://www.analyticsvidhya.com/blog/2020/03/6-python-libraries-interpret-machine-learning-models/ Depending on the project that you are working on, different model explainers will be able to tell you your results in different ways. There are 6 main model explainers, each with their own personality.
https://titanicexplainer.herokuapp.com/classifier/
resources: https://www.kaggle.com/code/dansbecker/advanced-uses-of-shap-values/tutorial https://medium.com/analytics-vidhya/explain-ml-models-shap-library-5ce375c85d7d https://www.kaggle.com/code/scratchpad/notebook616777f210/edit https://spark.apache.org/docs/latest/api/python/reference/api/pyspark.mllib.tree.GradientBoostedTrees.html https://interpret.ml/docs/lime.html https://github.com/TeamHG-Memex/eli5