SOFTWARE
I've created the following open source tools and packages:
StructureBoost: This is a package for gradient boosting that enables the algorithm to take into account the structure of categorical variables. Plus it has a bunch of other nifty features.
Github: https://www.github.com/numeristical/structureboost
Documentation: https://structureboost.readthedocs.io
SplineCalib: A tool for probability calibration using splines. The most advanced and feature-rich tool for effectively calibrating your model. SplineCalib is contained in the ML-Insights package.
ML-Insights: Along with SplineCalib, ML-Insights contains advanced ICE-plot visualization for understanding the performance of your model. It also contains the "histogram_pair" function for data exploration in classification problems.
Github: https://www.github.com/numeristical/introspective
Documentation: https://ml-insights.readthedocs.io