This project includes a Python implementation of the k-Nearest Neighbors (KNN) algorithm, a simple yet powerful algorithm used for classification and regression tasks in machine learning. The provided script (knn.py) contains functions to load data and predict labels for data points based on their nearest neighbors.
- Data Loading: Functions to load testing and training data from CSV files.
- KNN Classification: Implementation of the KNN algorithm using Euclidean distance to determine the closest neighbors.
- Tie Breaking: Mechanism to handle ties between potential labels for a classification.
- Python 3.x
- NumPy
- Matplotlib
To use this KNN implementation, clone the repository to your local machine:
git clone https://github.com/megamiii/KNN.gitThe main script can be run in a Python environment with the necessary libraries installed. Ensure you have the dataset in the correct format as expected by the load_knn_data function.
For an in-depth analysis of the KNN algorithm's performance, including observations from output graphs and terminal results, please see the RESULTS.md file in this repository. It provides a comprehensive overview of the algorithm's accuracy, optimal parameters, and generalization capabilities based on the conducted tests.
README.md: Provides an overview of the project, including how to set up and run the KNN algorithm.RESULTS.md: Contains detailed analysis of the KNN algorithm's performance, the output graph, and the terminal output summary.knn.py: The main script with KNN algorithm implementation and data loading functions.knn-dataset/:train_inputs.csv: Training data input features.train_labels.csv: Training data labels.test_inputs.csv: Test data input features.test_labels.csv: Test data labels.
results/:result.png: The graph showing the relationship between the number of neighbors and accuracy.output.png: The terminal output showcasing the results of the KNN algorithm.