purnamatools 0.2.2

Python package to simplify the initial stages of model building and analysis for data science projects

PurnamaTools

PurnamaTools is a Python package designed to simplify data analysis and machine learning workflows, especially for beginners. It provides utilities for:

• Initial data inspection and validation
• Feature selection and correlation analysis
• Model evaluation for regression and classification

The package is intended to be continuously updated with new features and improvements.

Features

1. Initial Data Check

• initial_data_overview(df, target=None, is_classification=True): Comprehensive overview of your dataset, including missing values, duplicates, outliers, low variance columns, and more.
• check_class_balance(df, target): Quickly check class imbalance and get recommendations for classification tasks.

2. Feature Selection

• correlation_analysis(df, target, method='pearson'): Identify strong correlations and potential redundant features.
• mi_analysis(X, y): Select top features based on Mutual Information.
• batch_rfe_feature_selection(X, y): Scalable Recursive Feature Elimination for datasets with many features.
• sfs_feature_selection(X, y): Sequential Feature Selection (forward or backward) using any estimator.
• lasso_feature_selection(X, y): Feature selection using Lasso regression.

3. Model Evaluation

• evaluate_model_regression(model, X_train, y_train, X_test, y_test, scoring): Evaluate regression models and get guidance for overfitting or underfitting.
• evaluate_model_classification(model, X_train, y_train, X_test, y_test, scoring): Evaluate classification models and get suggestions for handling class imbalance.

Installation

pip install purnamatools