ML-MyKNearestNeighbors

Overview

The MyKNN.py file contain my home-made version of the k-Nearest-Neighbors classifier.

By default k = 5, it means that the classifier will checks for the closest 5 neightbors of the data you want to test.

Dependencies

The classifier only uses basic library (math, sys and operator), so you don't need dependencies to use it.

The main file (main.py) uses sklearn to get datasets, but you don't need it if you want to use just the classifier. Run this command to get sklearn :

pip install scikit-learn

Run the script

You can run this script in terminal with this command line :

python main.py

Use the classifier

First create the classifier.

from MyKNN import *
classifier = MyKNN()

# You can change the value of k by this
classifier.k = 5

Then, train it.

# x_train is an array of features (like [[1, 2], [5, 9], [6, 8], [2, 3]])
# y_train is an array of labels   (like [  'a' ,  'b'  ,  'b'  ,  'a'  ])
# Labels index must match the corresponding feature index
classifier.train(x_train, y_train)

And now, you can predict some output.

# x_test is an array of features you want to get the label (like [[6, 8], [0, 2]])
predictions = classifier.predict(x_test)
print(predictions)
# Display : ['b', 'a']