Rで学ぶクラスタ解析のコードをpythonで書いただけ

mydefs.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

# Rで学ぶクラスタ解析のパクり

import scipy as sp


def table(result, answer):
    result, answer = sp.asarray(result), sp.asarray(answer)
    ret = []
    for i in sorted(set(result)):
        tmp = []
        for j in sorted(set(answer)):
            tmp.append(len(answer[result==i][answer[result==i]==j]))
        ret.append(tmp)
    return sp.array(ret)


def mypurity(ct):
    return sp.sum(sp.apply_along_axis(max, 1, ct)) / float(sp.sum(ct))


def myentropy(ct):
    ct = sp.array(ct, dtype=float) # 面倒なので全部floatに;
    def _entropy0(pv):
        p1 = pv / sp.sum(pv)
        p2 = p1[p1 != 0]
        return sp.sum(p2 * sp.log(p2))
    return -sp.sum((sp.sum(ct, 1) / sp.sum(ct)) * sp.apply_along_axis(_entropy0, 1, ct)) / sp.log(ct.shape[1])


def myeval(ans, goldans):
    ct = table(ans, goldans)
    print "Entropy: ", myentropy(ct)
    print "Purity: ", mypurity(ct)