overlapping.py

import numpy as np

def volume_overlap(X, y):
    classes=np.unique(y)
    totalvoverlap=0
    counter=0
    r, c = np.shape(np.array(X))
    for v0 in range(len(classes)):
        for v1 in range(len(classes)):
            if (classes[v0] != classes[v1] and classes[v0]<classes[v1]):
                counter=counter+1
                voverlap = 1.0
                for i in range(c):
                    max0 = -100000.0
                    max1 = -100000.0
                    min0 = 100000.0
                    min1 = 100000.0

                    for j in range(r):
                        if (y[j] == classes[v0]):
                            if (max0 < X[j, i]):
                                max0 = X[j, i]

                            if (min0 > X[j, i]):
                                min0 = X[j, i]

                        if (y[j] == classes[v1]):
                            if (max1 < X[j, i]):
                                max1 = X[j, i]

                            if (min1 > X[j, i]):
                                min1 = X[j, i]
                    if (np.maximum(max0, max1) - np.minimum(min0, min1)==0):
                        voverlap=0
                    else:
                        voverlap = voverlap * ((np.minimum(max0, max1) - np.maximum(min0, min1)) / (np.maximum(max0, max1) - np.minimum(min0, min1)))
                totalvoverlap=totalvoverlap+voverlap
    avgvoverlap=totalvoverlap/counter

    return avgvoverlap