biserial-test.sagews

import scipy.stats
f = open('beauty.csv', 'r')
head = f.readline()
data = f.readlines()
f.close()
for k in xrange(2, 8) :
    wage = []
    x = []
    for row in data :
        p = row.split(';')
        wage.append(float(p[0]))
        x.append(int(p[k]))
    wage, x = vector(wage), vector(x)
    rb = scipy.stats.pointbiserialr(x, wage)
    print k, rb.correlation, rb.pvalue

2 0.094529221781 0.000780413903579 3 0.00675565935638 0.810666070611 4 -0.0590517302681 0.0360930890199 5 -0.313419441587 4.0503960453e-30 6 0.184520769104 4.10912116992e-11 7 -0.0536527693027 0.0569146655792