lukakostic / kmeansclustering-knearestneighbour

Minimal implementation of KMeansClustering and KNearestNeighbour classifiers (simple machine learning)

KMeansClustering and KNearestNeighbour

Minimal implementation of KMeansClustering and KNearestNeighbour classifiers (simple machine learning), in C#

Fairly simple and commented, check KMeansClustering() and KNearestNeighbour() methods.

Nearest Neighbour - place each element into n-dimensional space (dimension based on feature count), then when you need to classify a custom element, you also place it in n-dimensional space and find the closest one with euclidian distance (pythagoras). Can also be used for regression.

Clustering - same as Nearest Neighbour but you form clusters for each type and find closest cluster center like you did for nearest Neighbour

Has the sample trainging data i made up, as shown in picture(s).

Clustering has no much use here as i have only 1 element per type, but if you have a better dataset then it can shine

Both predict the type/class of an element based on its features

kmeansclustering-knearestneighbour's Issues

Images