StackerPy

Model Stacking for Scikit-Learn Models (including the ability to blend)

More Details on Website @: StackerPy - Model Stacking For Scikit-Learn Models

Tech

StackerPy uses a number of open source projects to work properly:

• SciKit-Learn
• Numpy
• Pandas
• Matplotlib

And of course StackerPy itself is open source with a public repository on GitHub.

Installation

Install the dependencies (although pip should do this when installing stackerpy)

$ pip install numpy
$ pip install pandas
$ pip install matplotlib
$ pip install sklearn

Install the package

$ pip install stackerpy

How to use

Stacker Model:

# base models

import pandas as pd
import numpy as np

from sklearn.linear_model import LogisticRegression, RidgeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.tree import DecisionTreeClassifier
from stackerpy import StackerModel

from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split

data = load_breast_cancer()

X = pd.DataFrame(data.data)
y = pd.DataFrame(data.target)

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=np.sum([ord(i) for i in 'StackerPy'])
)

lr2 = LogisticRegression(solver='lbfgs')
dt2 = DecisionTreeClassifier()
rf2 = RandomForestClassifier()
models = [lr2, dt2, rf2]

# stacker
rc2 = RidgeClassifier()

#fitting
stacker = StackerModel()
stacker.fit(
    X=X_train,
    y=y_train,
    models=models,
    stacker=rc2,
    blend=True,
    splits=5,
    model_feature_indices=None)

# predicting
stacker_predictions = stacker.predict(X_test)

Performance

License

MIT