learnify-ml 0.0.22

A package for auto preprocessing and training machine learning models

learnify-ml

learnify-ml is a lightweight and modular Python package designed to automate preprocessing pipelines for machine learning datasets. With minimal configuration, it applies standard best practices to clean, transform, and prepare data for training.

---

Features

Data Preprocessing

• Automatically detects and fills missing values
• Identifies and handles outliers
• Applies skewness correction to numeric features
• Scales numerical data using standard scaling
• Encodes categorical variables using label encoding
• Removes low-variance features
• Performs VIF analysis to reduce multicollinearity
• Balances imbalanced datasets using SMOTE
• Supports feature selection methods

Model Training & Evaluation

• Trains multiple machine learning models (e.g., Random Forest, XGBoost, Logistic Regression)
• Automatically selects the best-performing model based on scoring metrics
• Tracks performance with metrics like accuracy, F1-score
• Saves trained model and evaluation report as artifacts

Hyperparameter Optimization

• Supports both GridSearchCV and RandomizedSearchCV
• Configurable search space for each model
• Automatically selects best hyperparameters and retrains final model

---

Installation

pip install learnify-ml

If the package is not yet on PyPI, you can install locally:

git clone https://github.com/yourusername/learnify-ml.git
cd learnify-ml
pip install -e .

Usage

🔹 1. Import the main pipeline class

from learnify_ml import AutoMLPipeline

🔹 2. Run the pipeline

trainer = AutoMLPipeline(target_column="target_column",
                      use_case="regression",
                      apply_hyperparameter_tuning=True,
                      hyperparameter_tuning_method="randomized",
                      apply_tf_idf=False,
                      apply_scale=True,
                      apply_feature_selection=True,
                      apply_outlier=True,
                      apply_vif=True,
                      apply_skewness=True,
                      apply_smote=False,
                      test_size=0.2,
                      impute_strategy="mean",
                      ).run_pipeline()
trainer.run_pipeline()

Example Functions

df = trainer.remove_outliers(df)
df = trainer.handle_missing_values(df)
df = trainer.scale_numeric_features(df)

Each method is modular and can be used independently or inside run_pipeline().