stacking 0.1.1

Tree of files

• base_fixed_fold.py (class of stacking)

• data/

• input/

  • train.csv (train dataset)

  • test.csv (test dataset)

• output/

  • features/

  • features.csv (features user created)

  • temp/

  • temp.csv (files saved in stacking)

• scripts/

• script.csv (main script where concrete models defined)

Installation

python setup.py install

Details of scripts

• base.py: Base models for stacking are defined here (using sklearn.base.BaseEstimator). Some models are defined here. e.g., XGBoost, Keras, Vowpal Wabbit. These models are wrapped as scikit-learn like (using sklearn.base.ClassifierMixin, sklearn.base.RegressorMixin). That is, model class has some methods, fit() and predict_proba().

New user-defined models can be added here.

Scikit-learn models can be used.

Base model have some arguments. ‘s’: Stacking. Svaing a oof prediction({model_name}_all_fold.csv) and average of test prediction based on fold-train models({model_name}_test.csv). These files will be used for next level stacking. ‘t’: Training with all data and predict test({model_name}_TestInAllTrainingData.csv). This is useful to get the single model performance. ‘st’: Stacking and then training with all data and predict test (‘s’ and ‘t’). ‘cv’: Only cross validation without saving the prediction.

Define task details top of script.

• features.py: Create features based on original dataset.

• scripts/XXX.py: Define several models and its parameters used for stacking. Train and test feature set are defined here. Need to define CV-fold index.

Any level stacking can be defined.

TODO LIST

Need to be more general library.

Please check isuues!!