SkOpts 0.0.1

Hyperparameters tuning of machine learning models provided by sklearn library using optuna

SkOpts

Library providing high-level functions for easy hyperparameters tuning of models using Optuna.

This repository contains the project of a library that offers ready-to-use functions to optimize the hyperparameters of some of models provided by the Sklearn API, XGBoost, LightGBM and CatBoost.

Code snippet to optimize a classification model:

from Tune_Xgboost import XGB_tuner

XGB_tuned=XGB_tuner(X=Xtrain,y=y_train,
                    scoring_metric='roc_auc',
                    n_trials=100,
                    N_folds=5,
                    direction='maximize',
                    stratify=True,
                    problem_type='classification')

Code snippet to optimize regression model:

from Tune_Xgboost import XGB_tuner

XGB_tuned=XGB_tuner(X=Xtrain,y=y_train,
                    scoring_metric='neg_mean_squared_error',
                    n_trials=100,
                    N_folds=5,
                    direction='minimize',
                    problem_type='regression')

Parameter	Usage
"X"	Training dataset without target variable.
"y"	Target variable.
"scoring_metric"	Metric to optimize.
"n_trials"	Number of trials to execute optimization.
"N_folds"	Number of folds for cross validation.
"direction"	Equals "maximize" or "minimize".
"problem_type"	Equals "classification" or "regression".
"stratify"	Stratify cv splits based on target distribuition [True or False]

The 'scoring_metric' parameter takes the same values from sklearn API (link of available list: https://scikit-learn.org/stable/modules/model_evaluation.html)