ML Algorithms from Scratch

Linear Regression, Logistic Regression, Decision Tree, KNN, and Naive Bayes implemented from scratch using NumPy — MSDS Coursework

MSDS Coursework Batangas State University 2025–2026

GitHub Repository →

Overview

Implemented 5 classical ML algorithms entirely from first principles using NumPy — no sklearn for the core logic, just math and manual formula implementation. Each algorithm is paired with a step-by-step written explanation, dataset, and output figures.

Algorithms Implemented

Algorithm Dataset Key Result
Linear Regression TV Marketing (200 samples) Pearson’s r = 0.782
Logistic Regression Pima Indians Diabetes (768 samples) Sigmoid probability visualisation
Decision Tree Iris (150 samples) Train: 78.57% / Test: 71.05%
K-Nearest Neighbors Pima Indians Diabetes (768 samples) Best k=5, Accuracy: 72.73%
Naive Bayes Iris (150 samples) Accuracy: 90.00%

What Each Implementation Covers

  • Formula derivation and manual step-by-step computation
  • No sklearn for core algorithm logic — only for data splitting and evaluation metrics
  • Written explanation (Markdown + PDF) with figures for each algorithm
  • Validation on real-world datasets

Tech Stack: Python · NumPy · Pandas · Matplotlib · Seaborn · Google Colab