gzoo.py 文件源码

def measure_correlation(gold0, gold1, l = 10, pos = 0, list_w = []):

    a = []
    b = []

    for w in gold0.keys():
      if w in gold1 and (w in list_w or list_w == []):
        if gold0[w][0] != None and gold1[w][0] != None:
          if gold0[w][2] > l and gold1[w][2] > l:
              if pos == 0:
                  a.append(logit(gold0[w][pos]))
                  b.append(logit(gold1[w][pos]))
              else:
                  a.append(logit(1 - gold0[w][pos]))
                  b.append(logit(1 - gold1[w][pos]))


    print len(a), np.mean(a), np.mean(b)
    return np.cov(a,b)