preprocess_module.py 文件源码

def generateStandardization(source_attr, target_attr, source_index):
    tempSum = 0.0
    anoSum = 0.0
    for each in source_attr:
        tempSum += each[source_index]
    avg = tempSum/float(len(source_attr))

    for each in source_attr:
        anoSum += float(math.pow((each[source_index] - avg), 2))
    standardV = anoSum/float(len(source_attr))
    standardV = math.sqrt(standardV)

    for i in range(len(source_attr)):
        temp = source_attr[i][source_index]
        res = (temp-avg) / standardV
        target_attr[i].append(res)

# Directly copy attr without modifiy