Source code I've used to produce this analysis.
Before running any code I recommend you to start a new virtual environment and install all the packages in requirements.txt.
To generate the dataset needed for the analysis and the ML model you first need to run these steps one by one
1 - Download data download_data.py
2 - Decompress data with decompress_data.py
3 - Extract headers with extract_headers.py
4 - Store data in a single parquet with compact_all_data.py
You can also download directly the dataset from Kaggle (link), however this dataset is not updated. If you want to use an up to date dataset I recommend to run above scripts.
The analysis is in the notebooks/AB-testing.ipynb notebook.
The ML model and its evaluation are in the notebooks/ML.ipynb notebook.