Recently, I completed the development of a Python package for the k-nearest neighbors (k-NN) classifier. This package is available on the Python Package Index (PyPI) and GitHub (links at the bottom of this page). In this post, I will give an overview of the algorithm and the usage of my package.
The algorithm
The k-nearest neighbors classifier is a supervised learning algorithm used for classification tasks. It works by finding the k closest data points in the feature space to a given input data point. The class label of the majority of these nearest neighbors is assigned to the input data point, making it a non-parametric and instance-based learning method. The value of k, which represents the number of neighbors to consider, is a hyperparameter that influences the model’s performance and can be tuned based on the dataset and problem at hand.
Here’s a visual representation:
Imagine our data points can be represented in two-dimensional space. Every data point has a label. Since the value of k is 3, the algorithm considers the labels of the three nearest neighbors to the input data point and predicts the label of the input based on the majority of these neighbors. In this case, the algorithm would predict a label of 1.
The package
For an in-depth walkthrough of each step of using this package, check out the GitHub page (linked at the bottom). Here is an overview of the steps:
- Prepare your data: Have your training and testing data ready in text files or arrays.
- Fit the classifier: Initialize a classifier object and fit it to your training data.
- Make predictions: Compute and store label predictions for your testing data. This is where you specify the k value.
- Check accuracy: Evaluate the accuracy of your label predictions against the actual labels of the testing data.
This package can execute predictions quickly on large datasets and is flexible with the format and organization of the data, thanks to NumPy arrays and vector norm methods.
Justin Roche 
Leave a comment