mcbarlowe / lrboost

LRBoost is a scikit-learn compatible approach to performing linear residual based stacking/boosting.

|ReadTheDocs|_

LRBoost - Linear Residual Boosting with scikit-learn

LRBoost is a combination of a linear estimator and gradient boosting that is scikit-learn compatible.

• LRBoostRegressor can be used like any other sklearn estimator and is built off a sklearn template.

• predict returns an array-like of final predictions

• predict_detail returns a dictionary with the linear, non-linear, and final predictions.

• predict(X) and predict_detail(X)['pred'] are equivalent values

  >>> from sklearn.datasets import load_iris
  >>> from lrboost import LRBoostRegressor
  >>> X, y = load_iris(return_X_y=True)
  >>> lrb = LRBoostRegressor.fit(X, y)
  >>> predictions = lrb.predict(X)
  >>> detailed_predictions = lrb.predict_detail(X)

More detailed documentation can be found here -> https://readthedocs.org/projects/lrboost.

Andrew Patton & Kostya Medvedovsky