yay.py

 while norm > epsilon:
        iteration += 1
        norm = dist_method(prototypes, prototypes_old)
        prototypes_old = prototypes
        #for each instance in the dataset
        for index_instance, instance in enumerate(dataset):
            #define a distance vector of size k
            dist_vec = np.zeros((k, 1))
            #for each centroid
            for index_prototype, prototype in enumerate(prototypes):
                #compute the distance between each centroid and data point and store them
                dist_vec[index_prototype] = dist_method(prototype,
                                                        instance)
            #find the smallest distance, assign that distance to a cluster
            belongs_to[index_instance, 0] = np.argmin(dist_vec)
        
        tmp_prototypes = np.zeros((k, num_features))

        #for each cluster (k of them)
        for index in range(len(prototypes)):
            #get all the points assigned to a cluster
            instances_close = [i for i in range(len(belongs_to)) if belongs_to[i] == index]
            #find the mean of those points, this is our new centroid
            prototype = np.mean(dataset[instances_close], axis=0)
            #add our new centroid to our new temporary list
            tmp_prototypes[index, :] = prototype
        
        #set the new list to the current list
        prototypes = tmp_prototypes
        
        #add our calculated centroids to our history for plotting
        history_centroids.append(tmp_prototypes)

    #return calculated centroids, history of them all, and assignments for which cluster each datapoint belongs to
    return prototypes, history_centroids, belongs_to