myml-regression 0.1.0

A custom ML regression package built from scratch without Scikit-learn

myml-regression

A custom machine learning regression package built from scratch without Scikit-learn.

Features

Module	Functions / Classes
Regression	OLSRegression, RidgeRegression, LassoRegression
Preprocessing	fill_missing_mean, fill_missing_mean_df, normalize_features, add_intercept, detect_outliers_zscore
Feature Selection	forward_selection, backward_elimination
Diagnostics	normality_test (Shapiro-Wilk), vif (Variance Inflation Factor), heteroscedasticity
Visualization	plot_actual_vs_predicted, residual_plot
Prediction	predict_values

Installation

From PyPI (once published)

pip install myml-regression

From source

git clone https://github.com/yourusername/myml-regression.git
cd myml-regression
pip install .

From local wheel

pip install dist/myml_regression-0.1.0-py3-none-any.whl

Quick Start

import numpy as np
from myml.preprocessing import add_intercept, normalize_features, fill_missing_mean
from myml.regression import OLSRegression, RidgeRegression, LassoRegression
from myml.feature_selection import forward_selection
from myml.diagnostics import normality_test, vif

# --- Sample data ---
X = np.array([[1, 2], [3, 4], [5, 6], [7, 8]], dtype=float)
y = np.array([1.5, 3.5, 5.5, 7.5])

# Preprocessing
X_norm = normalize_features(X)
X_int  = add_intercept(X_norm)

# OLS Regression
ols = OLSRegression()
ols.fit(X_int, y)
print("OLS predictions:", ols.predict(X_int))

# Ridge Regression
ridge = RidgeRegression(alpha=0.5)
ridge.fit(X_int, y)
print("Ridge predictions:", ridge.predict(X_int))

# Lasso Regression
lasso = LassoRegression(alpha=0.01, iterations=1000)
lasso.fit(X_int, y)
print("Lasso predictions:", lasso.predict(X_int))

# Diagnostics
residuals = y - ols.predict(X_int)
stat, p = normality_test(residuals)
print(f"Shapiro-Wilk p-value: {p:.4f}")
vif_scores = vif(X_int)
print("VIF scores:", vif_scores)

Dependencies

• numpy >= 1.21
• pandas >= 1.3
• matplotlib >= 3.4
• scipy >= 1.7

Publishing to PyPI

# Install build tools
pip install build twine

# Build distributions
python -m build

# Upload to TestPyPI first
twine upload --repository testpypi dist/*

# Upload to PyPI
twine upload dist/*

Project Structure

myml_package/
├── myml/
│   ├── __init__.py
│   ├── regression.py        # OLS, Ridge, Lasso
│   ├── preprocessing.py     # Missing values, normalization, outliers
│   ├── feature_selection.py # Forward selection, backward elimination
│   ├── diagnostics.py       # Shapiro-Wilk, VIF, heteroscedasticity
│   ├── visualization.py     # Actual vs Predicted, Residual plot
│   └── prediction.py        # predict_values helper
├── tests/
│   └── test_myml.py
├── pyproject.toml
├── setup.py
├── setup.cfg
├── MANIFEST.in
├── LICENSE
└── README.md

License

MIT