modelLab 0.3.4

A lib for automating model training process of choosing best model that works for you data

modelLab is a comprehensive library of machine learning models designed to facilitate regression or classification tasks on a given dataset. It encompasses a diverse range of models and provides a comprehensive evaluation of each model’s performance, delivering a comprehensive set of metrics in a Python dictionary.

PURPOSE OF THE PACKAGE

• The primary objective of the package is to offer a curated ensemble of renowned scikit-learn models, enabling users to conveniently train all models with a single function call.

FEATURES

• Collections of Machine learning models

  • Classification Models

    • ‘LinearSVC’

    • ‘SGDClassifier’

    • ‘MLPClassifier’

    • ‘Perceptron’

    • ‘LogisticRegression’

    • ‘LogisticRegressionCV’

    • ‘SVC’

    • ‘CalibratedClassifierCV’

    • ‘PassiveAggressiveClassifier’

    • ‘LabelPropagation’

    • ‘LabelSpreading’

    • ‘RandomForestClassifier’

    • ‘GradientBoostingClassifier’

    • ‘QuadraticDiscriminantAnalysis’

    • ‘HistGradientBoostingClassifier’

    • ‘RidgeClassifierCV’

    • ‘RidgeClassifier’

    • ‘AdaBoostClassifier’

    • ‘ExtraTreesClassifier’

    • ‘KNeighborsClassifier’

    • ‘BaggingClassifier’

    • ‘BernoulliNB’

    • ‘LinearDiscriminantAnalysis’

    • ‘GaussianNB’

    • ‘NuSVC’

    • ‘DecisionTreeClassifier’

    • ‘NearestCentroid’

    • ‘ExtraTreeClassifier’

    • ‘DummyClassifier’

  • Regression Models

    • ‘SVR’

    • ‘RandomForestRegressor’

    • ‘ExtraTreesRegressor’

    • ‘AdaBoostRegressor’

    • ‘NuSVR’

    • ‘GradientBoostingRegressor’

    • ‘KNeighborsRegressor’

    • ‘HuberRegressor’

    • ‘RidgeCV’

    • ‘BayesianRidge’

    • ‘Ridge’

    • ‘LinearRegression’

    • ‘LarsCV’

    • ‘MLPRegressor’

    • ‘XGBRegressor’

    • ‘CatBoostRegressor’

    • ‘LGBMRegressor’

• Can also be used for the custom models.

GETTING STARTED

This package is available on PyPI, allowing for convenient installation through the PyPI repository.

Dependencies

-  'scikit-learn'
-  'xgboost'
-  'catboost'
-  'lightgbm'

INSTALLATION

If you already installed scikit-learn, the easiest way to install modelLab is using pip:

pip install modelLab

USAGE

>>> from modelLab import regressors,classifier
>>> regressors(X, y, models=models, verbose=False, rets=True) #X,y is data
>>> classifier(X, y, models=models, verbose=False, rets=True)

Examples

• Regression Problem

>>> from modelLab import regressors
>>> from sklearn.datasets import fetch_california_housing
>>> X,y=fetch_california_housing(return_X_y=True)
>>> regressors(X,y,verbose=True)
Model: SVR
Adjusted R^2: -0.0249
R^2: -0.0229
MSE: 1.3768
RMSE: 1.1734
MAE: 0.8698

Model: RandomForestRegressor
Adjusted R^2: 0.8034
R^2: 0.8038
MSE: 0.2641
RMSE: 0.5139
MAE: 0.3364

Model: ExtraTreesRegressor
Adjusted R^2: 0.8102
R^2: 0.8105
MSE: 0.2550
RMSE: 0.5050
MAE: 0.3333

Model: AdaBoostRegressor
Adjusted R^2: 0.4563
R^2: 0.4574
MSE: 0.7304
RMSE: 0.8546
MAE: 0.7296

Model: NuSVR
Adjusted R^2: 0.0069
R^2: 0.0088
MSE: 1.3342
RMSE: 1.1551
MAE: 0.8803

Model: GradientBoostingRegressor
Adjusted R^2: 0.7753
R^2: 0.7757
MSE: 0.3019
RMSE: 0.5494
MAE: 0.3789

Model: KNeighborsRegressor
Adjusted R^2: 0.1435
R^2: 0.1451
MSE: 1.1506
RMSE: 1.0727
MAE: 0.8183

Model: HuberRegressor
Adjusted R^2: 0.3702
R^2: 0.3714
MSE: 0.8461
RMSE: 0.9198
MAE: 0.5800

Model: RidgeCV
Adjusted R^2: 0.5868
R^2: 0.5876
MSE: 0.5551
RMSE: 0.7450
MAE: 0.5423

Model: BayesianRidge
Adjusted R^2: 0.5868
R^2: 0.5876
MSE: 0.5551
RMSE: 0.7451
MAE: 0.5422

Model: Ridge
Adjusted R^2: 0.5867
R^2: 0.5875
MSE: 0.5552
RMSE: 0.7451
MAE: 0.5422

Model: LinearRegression
Adjusted R^2: 0.5867
R^2: 0.5875
MSE: 0.5552
RMSE: 0.7451
MAE: 0.5422

Model: LarsCV
Adjusted R^2: 0.5211
R^2: 0.5220
MSE: 0.6433
RMSE: 0.8021
MAE: 0.5524

Model: MLPRegressor
Adjusted R^2: -3.5120
R^2: -3.5032
MSE: 6.0613
RMSE: 2.4620
MAE: 1.7951

Model: XGBRegressor
Adjusted R^2: 0.8269
R^2: 0.8272
MSE: 0.2326
RMSE: 0.4822
MAE: 0.3195

Model: CatBoostRegressor
Adjusted R^2: 0.8461
R^2: 0.8464
MSE: 0.2068
RMSE: 0.4547
MAE: 0.3005

Model: LGBMRegressor
Adjusted R^2: 0.8319
R^2: 0.8322
MSE: 0.2259
RMSE: 0.4753
MAE: 0.3185

• Classification Problem

>>> from modelLab import regressors,classifier
>>> from sklearn.datasets import load_iris
>>> X,y=load_iris(return_X_y=True)
>>> import warnings
>>> warnings.filterwarnings('ignore')
>>> classifier(X,y,verbose=True)
Model: LinearSVC
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: SGDClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9661

Model: MLPClassifier
Accuracy: 1.0000
Precision: 1.0000
Recall: 1.0000
F1 Score: 1.0000

Model: Perceptron
Accuracy: 0.8667
Precision: 0.9022
Recall: 0.8667
F1 Score: 0.8626

Model: LogisticRegression
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: SVC
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: CalibratedClassifierCV
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: PassiveAggressiveClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: LabelPropagation
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: LabelSpreading
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: RandomForestClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: GradientBoostingClassifier
Accuracy: 0.9333
Precision: 0.9436
Recall: 0.9333
F1 Score: 0.9331

Model: QuadraticDiscriminantAnalysis
Accuracy: 1.0000
Precision: 1.0000
Recall: 1.0000
F1 Score: 1.0000

Model: HistGradientBoostingClassifier
Accuracy: 0.9000
Precision: 0.9214
Recall: 0.9000
F1 Score: 0.8989

Model: RidgeClassifierCV
Accuracy: 0.8667
Precision: 0.8754
Recall: 0.8667
F1 Score: 0.8662

Model: RidgeClassifier
Accuracy: 0.8667
Precision: 0.8754
Recall: 0.8667
F1 Score: 0.8662

Model: AdaBoostClassifier
Accuracy: 0.9333
Precision: 0.9436
Recall: 0.9333
F1 Score: 0.9331

Model: ExtraTreesClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: KNeighborsClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: BaggingClassifier
Accuracy: 0.9333
Precision: 0.9436
Recall: 0.9333
F1 Score: 0.9331

Model: BernoulliNB
Accuracy: 0.2333
Precision: 0.0544
Recall: 0.2333
F1 Score: 0.0883

Model: LinearDiscriminantAnalysis
Accuracy: 1.0000
Precision: 1.0000
Recall: 1.0000
F1 Score: 1.0000

Model: GaussianNB
Accuracy: 0.9333
Precision: 0.9333
Recall: 0.9333
F1 Score: 0.9333

Model: NuSVC
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: DecisionTreeClassifier
Accuracy: 0.9333
Precision: 0.9436
Recall: 0.9333
F1 Score: 0.9331

Model: NearestCentroid
Accuracy: 0.9000
Precision: 0.9025
Recall: 0.9000
F1 Score: 0.9000

Model: ExtraTreeClassifier
Accuracy: 0.9667
Precision: 0.9694
Recall: 0.9667
F1 Score: 0.9667

Model: DummyClassifier
Accuracy: 0.2333
Precision: 0.0544
Recall: 0.2333
F1 Score: 0.0883

• Using Custom Models

>>> from sklearn.datasets import make_regression
>>> from sklearn.linear_model import LinearRegression
>>> from modelLab import regressors
>>> X, y = make_regression(n_samples=100, n_features=10, random_state=42)
>>> models = {'Linear Regression': LinearRegression()}
>>> regressors(X, y, models=models, verbose=False, rets=True)
defaultdict(<class 'dict'>, {'Linear Regression': {'Adjusted R^2': 1.0, 'R^2': 1.0, 'MSE': 3.097635893749451e-26, 'RMSE': 1.7600101970583725e-13, 'MAE': 1.4992451724538115e-13}})

>>> from sklearn.datasets import make_regression, make_classification
>>> from sklearn.linear_model import LogisticRegression
>>> from modelLab import classifier
>>> X, y = make_classification(n_samples=100, n_features=10, random_state=42)
>>> models = {'Logistic Regression': LogisticRegression()}
>>> classifier(X, y, models=models, verbose=False, rets=True)
defaultdict(<class 'dict'>, {'Logistic Regression': {'Accuracy': 0.95, 'Precision': 0.9545454545454545, 'Recall': 0.95, 'F1 Score': 0.949874686716792}})

Contributor and Author

[Abhishek Kaddipudi]

Mail

Linkedin

GitHub