100 Add modes: `Explain`, `Perform`, `Compete` 108 Faster NN implementation for small datasets 110 Fix warnings in permutation importance
0.4.1
0.4.0
Enhancements - 96 Use internal early stopping - 95 convergence warning in linear algorithm - 94 Add support for kNN - 90 Try to provide more meaningful names to models - 74 Add stacked models - 72 Add support for Neural Networks - 70 Select best model after each iteration - 64 Generate data information once - 50 Add validation with split
Bugs - 91 Dont run preprocessing for baseline algorithm - 85 Cant load CatBoost model for predictions
0.3.5
Removed `mae` from sklearn decision tree based algorithms because they slow down the training
0.3.0
enhancement - 83 Compare all models visually - 79 Aggregate importances to one plot - 78 Sort linear model coefficients - 77 Shuffle generated models - 75 Add SHAP explanations to models - 69 Add `skip_interpret` argument - 61 Add ExtraTrees - 52 Add Linear and Logistic Regression support - 51 Add LightGBM support - 27 feature importance [enhancement] - 68 Add minimum number of steps for algorithm
bug - 66 Doesnt work with bytes as class - 65 Dont generate parameters for Baseline - 63 wrong time estimation for model training - 62 Hill climbing not iterating on all algorithms - 60 Show number of trees in learning curves