KNN

Files:

• /knn - module KNN for python.

• /tests - tests for KNN module.

• MNIST_KNN.ipynb - Experements with different classification algoritms on mnist datasets.

1. Exploration of the speed of algorithms on different dimensions of the feature space.
2. Exploration of the accuracy KNN + cross validation. Use two metrics: euclidean and cosine
3. Comparison weighted k nearest neighbors method and uniform
4. Overview of the method, comparison of the accuracy of standard dataset partitioning and cross-validation.Visualization (function plot_confusion_matrix in file: "plot_cm.py") and analysis of the confusion matrix.