K-Nearest Neighbors (KNN) Classifier from Scratch

A from-scratch implementation of K-Nearest Neighbors algorithm for classification, demonstrated on the classic Iris dataset.

Overview

This project includes:

• Complete KNN implementation with Euclidean, Manhattan, and Cosine distance metrics
• Data visualization using Seaborn and Matplotlib
• Model evaluation with accuracy scoring
• Clean, production-ready code with input validation

Features

• 🧮 Three distance metrics supported:
  • Euclidean distance
  • Manhattan distance
  • Cosine similarity (properly normalized)
• 📊 Interactive visualizations of dataset features
• 🔍 Comprehensive data exploration
• ⚙️ Configurable hyperparameters:
  • k-value (number of neighbors)
  • Task type (classification/regression)
  • Distance metric