/Cutstom-kNN-Classifier

Implementation of brute-force kNN classifier, testing kNN with wine dataset

Primary LanguageJupyter NotebookGNU General Public License v3.0GPL-3.0

k-Nearest Neighbors Classification

(Brute-force) k-Nearest Neighbors Classifier is implemented and used in wine-white dataset. All the required helper functions and performance metrics are also implemented. This custom implementation is cross checked with scikit learn.

  • knn_classifier.py: KNeighborsClassifier class. Methods: fit, predict, predict_proba
  • helper_functions.py: train_test_split, scale_normal - feature normalization.
  • metrics.py: Performance metrics functions - accuracy_score, precision_score, recall, f1-score, roc_curve, precision_recall etc.
  • distances.py: manhattan_distance, euclidean_distance etc.
  • cross_validation.py: kfold_cross_validation.
  • testing_knn_model.py: Using the custom implementation, we explore the wine-white dataset.