bla.py

def mean(C):
    m = np.array([0,0])
    for c in C:
        m += c
    m = m * (1/len(C))
    return m

def recalc(C, F, k, centroids):
    sqe = 0
    # assign features to clusters based on distance to centroids
    C = {}
    C[0] = list()
    C[1] = list()
    for f in F:
        #if np.linalg.norm((f-centroids[1])) < np.linalg.norm((f-centroids[0])):
        if distance.euclidean(f,centroids[1])**2 <= distance.euclidean(f,centroids[0])**2:
            C[1].append(f)
        else:
            C[0].append(f)
    print("----------------------------------------------------------------------")
    print("C",C)
    print("----------------------------------------------------------------------")
    # recalculate centroids
    centroids = [np.array([0,0]),np.array([0,0])]
    for i in range(k):
        #centroids[i] = np.mean(C[i], axis=0)
        centroids[i] = mean(C[i])
    for i in range(k):
        for c in C[i]:
            sqe += np.linalg.norm((c-centroids[i]))**2
            print("iter",i,"error", sqe)
    return (sqe, C, centroids)

def kmeans(F, k, centroids):
    sqe = 0
    psqe = None
    # create empty clusters
    C = {}
    print(C)
    #sqe, C = recalc(C, F, k, centroids)
    i = 0
    while sqe != psqe:
    #while i < 3:
        i += 1
        psqe = sqe
        sqe, C, centroids = recalc(C, F, k, centroids)
        #print(sqe, centroids, C)
    #print("mean", mean([np.array([1, 1]),np.array([5, 2])]))
    #print("mean", mean([np.array([1, 2]),np.array([2, 5]),np.array([4, 4]), np.array([2, 2]), np.array([5, 5])]))
    return (i, sqe, "CLUSTER", C, "c", centroids)


D = np.array([(1,1), (1,2), (2,2), (5,2), (4,4), (2,5), (5,5)])
    #print(find_centers(D, 2, np.array([D[3], D[4]]), np.array([D[1], D[2]])))
    c = np.array([D[3], D[4]])
    print("centroids", c)
    kms = kmeans(D, 2, c)
    print(kms)