Iglewicz and Hoaglin outlier test (modified Z-score test)

iglewicz_hoaglin.py

# based on:
#  * https://www.itl.nist.gov/div898/handbook/eda/section3/eda35h.htm
#  * http://colingorrie.github.io/outlier-detection.html#modified-z-score-method
#  * http://contchart.com/outliers.aspx
#  * https://stats.stackexchange.com/questions/339932/iglewicz-and-hoaglin-outlier-test-with-modified-z-scores-what-should-i-do-if-t

def calc_median(data):
    result = 0.0
    length = len(data)

    if length % 2 == 0:
        index01 = (length/2) - 1
        index02 = index01 + 1
        result = (data[index01] + data[index02])/2.0

    else:
        result = data[length//2]

    return result

def calc_mad(data, median):
    medians = []

    for x in data:
        medians.append(abs(x - median))

    medians.sort()
    mad =  calc_median(medians)

    if mad == 0:
        mad = 2.2250738585072014e-308 # sys.float_info.min

    return mad

def iglewicz_hoaglin(threshold, data):
    data.sort()
    median = calc_median(data)
    mad = calc_mad(data, median)
    result = []

    print('threshold:' + str(threshold))
    print('median:' + str(median))
    print('MAD:' + str(mad))
    print('data:'+ str(data))
    print('')

    for x in data:
    	score = abs(0.6745 * (x - median) / mad)
    	print(str(x) + ':\t' + str(score))

    	if(score > threshold):
    		result.append(x)
    
    return result

threshold = 3.5
data = [10, 22, 30, 100, 15, 80, 8, 9]

modified_z_scores = iglewicz_hoaglin(threshold,data)
print('\nresult: '+ str(modified_z_scores))