🏷️ KNN Classifier - IRIS Dataset

🚀 Project Overview

This project implements a K-Nearest Neighbors (KNN) classifier using Python and Scikit-learn. It focuses on the Iris dataset and demonstrates the full workflow of training, evaluating, and visualizing KNN models.

Key features include:

• Data preprocessing
• Model training with varying K values
• Accuracy evaluation
• Confusion matrix visualization
• Decision boundary visualization
• Cross-validation analysis including normal and weighted KNN models
• Comparison of results to identify the best-performing model

---

🛠️ Getting Started

1️⃣ Clone the repository

git clone https://github.com/gauthamdv/knn-classifier.git
cd knn-classifier

2️⃣ Install requirements

pip install -r requirements.txt

3️⃣ Prepare the Data

Run the preprocessing script to clean and prepare the dataset stored in the datasets/ folder:

python data-preprocessing.py

4️⃣ Run the Notebook

Open knn-classify.ipynb to:

• Train KNN models with different K values

• Evaluate performance using accuracy and confusion matrices

• Visualize decision boundaries

• Perform cross-validation analysis (normal and weighted)

• Compare results

• Identify the best-performing model based on the analysis (in this case, the regular weighted KNN model)

• Data: datasets/

• Models saved in: models/

• Plots saved in: plots/

---

📁 File Structure

knn-classifier/
├── datasets/                  # Raw and processed datasets
├── models/                    # Saved models
├── plots/                     # Visualizations and plots
├── .gitignore                 # Git ignore file
├── README.md                  # Project documentation (this file)
├── data-preprocessing.py      # Data preprocessing script
├── knn-classify.ipynb         # Jupyter notebook for KNN classification
└── requirements.txt           # Python dependencies

---

📊 Models & Evaluation

• KNN Classifier: Trained with multiple K values to find the optimal K.
• Cross-Validation: Evaluates normal and weighted KNN models and compares performance.
• Best Model Selection: After analyzing the comparison, the regular weighted KNN model was chosen as the best-performing model.
• Evaluation Metrics: Uses accuracy and confusion matrices.
• Visualization: Decision boundaries plotted for better understanding of model performance.

---

⚡ Usage

1. Prepare the data using data-preprocessing.py.
2. Open and run knn-classify.ipynb to train, evaluate, and cross-validate KNN models.
3. Compare results and select the best-performing model.
4. Check saved plots and models in their respective folders.

---

🤝 Contributions

Contributions are welcome! Feel free to open a pull request or report issues.

---