MLLearner 0.1.0 - Initial Release

MLLearner 0.1.0 - Initial Release

Overview

MLLearner is a comprehensive, educational Python package designed to help machine learning learners and practitioners understand and implement ML algorithms from scratch.

Features

📊 Data Handling Module

• DataLoader: Load CSV and JSON files
• DataProcessor: Handle missing values, encoding, normalization, standardization
• TrainTestSplit: Split datasets into train and test sets

🤖 ML Models

• Linear Regression: Gradient descent-based regression
• Logistic Regression: Binary classification with sigmoid activation
• K-Nearest Neighbors: KNN classifier with Euclidean distance
• Decision Trees: Tree-based classifier with Gini impurity

📈 Evaluation Metrics

• ClassificationMetrics: Accuracy, Precision, Recall, F1 Score, Confusion Matrix
• RegressionMetrics: MSE, RMSE, MAE, R² Score

🎓 Training Utilities

• Trainer: Base trainer for model training
• EarlyStoppingCallback: Prevent overfitting
• LearningRateScheduler: Adaptive learning rate scheduling

📊 Visualization Tools

• PlotUtils: Visualization utilities for metrics and results

🔧 Utility Functions

• ModelSaver: Save and load models
• DataValidator: Data validation
• HyperparameterTuner: Grid search for hyperparameters

Installation

pip install mllearner

Quick Start

from mllearner.data import DataLoader, DataProcessor, TrainTestSplit
from mllearner.models import LinearRegression
from mllearner.evaluation import RegressionMetrics

# Load and preprocess data
loader = DataLoader()
data = loader.load_csv('data.csv')

processor = DataProcessor()
X = processor.normalize(X)
y = processor.normalize(y)

# Split data
X_train, X_test, y_train, y_test = TrainTestSplit.split(X, y, test_size=0.2)

# Train model
model = LinearRegression(learning_rate=0.01, iterations=1000)
model.fit(X_train, y_train)

# Evaluate
y_pred = model.predict(X_test)
rmse = RegressionMetrics.rmse(y_test, y_pred)
print(f"RMSE: {rmse:.4f}")

Documentation

Full documentation available at: https://github.com/HarshTambade/ML-Learning-Hub

License

MIT License

Author

Harsh Tambade