import csv
import numpy as np
import random
from math import sqrt
import scipy.stats


def get_data(file):
    csv_x = csv.reader(open(file))
    data = []
    for row in csv_x:
        data.append(row)
    return data

def extrapolate(data_list, diagnosis):
    data = []
    for tuple in data_list:
        if tuple[0] == diagnosis:
            data.append(float(tuple[1]))
    return data

def sig_test(reference, symptom, tail):

    alpha = 0.05

    ref_sample_size = len(reference)
    sample_size = len(symptom)
    se = sqrt(np.std(symptom)**2/sample_size + np.std(reference)**2/ref_sample_size)
    t = abs(np.mean(symptom) - np.mean(reference))/se
    p_value = 1 - scipy.stats.t.cdf(t, min(sample_size, ref_sample_size) - 1)
    s = sqrt(((sample_size - 1)*np.std(symptom)**2 + (ref_sample_size - 1)*np.std(reference)**2)/(sample_size + ref_sample_size - 2))
    d = abs(np.mean(symptom) - np.mean(reference))/s
    if tail == 2:
        p_value *= 2

    print "   Standard error:", se, "   T-score:", t, "   P-value:", p_value, "   Difference of means:", np.mean(symptom) - np.mean(reference)

    if p_value <= alpha/3:
        print "Significant difference"
        print "Cohen's d:", d
        print "Difference of means:", np.mean(symptom) - np.mean(reference)
    else:
        print "No significant difference"

###########################################################

fever = get_data('fever.csv')
WBC = get_data('WBC.csv')
fever_None = extrapolate(fever, 'No')
fever_Py2_7 = extrapolate(fever, 'Py2.7')
fever_Py3 = extrapolate(fever, 'Py3')
WBC_None = extrapolate(WBC, 'No')
WBC_Py2_7 = extrapolate(WBC, 'Py2.7')
WBC_Py3 = extrapolate(WBC, 'Py3')

print "Fever tests:"
print "Normal and Py2.7:"
sig_test(fever_None, fever_Py2_7, 1)
print "Normal and Py3:"
sig_test(fever_None, fever_Py3, 1)
print "Py2.7 and Py3:"
sig_test(fever_Py2_7, fever_Py3, 2)
print ""
print "WBC tests:"
print "Normal and Py2.7:"
sig_test(WBC_None, WBC_Py2_7, 1)
print "Normal and Py3:"
sig_test(WBC_None, WBC_Py3, 1)
print "Py2.7 and Py3:"
sig_test(WBC_Py2_7, WBC_Py3, 2)