topsis.py

import pandas as pd
import numpy as np
import sys
import math as m


def main():

    if len(sys.argv) < 5:
        print('NOT ENOUGH ARG')
        sys.exit()

    fileName = sys.argv[1]
    weightage = sys.argv[2]
    weightage = weightage.split(",")
    weightage = [float(i) for i in weightage]
    # print(weightage)

    impact = sys.argv[3]
    impact = impact.split(",")
    # print(impact)

    resultFile = sys.argv[4]
    # print(resultFile)

    dataset = pd.read_csv(fileName)
    # print(dataset)
    data = dataset.iloc[:, 1:].values.astype(float)

    (r, c) = data.shape

    s = sum(weightage)

    if len(weightage) != c:
        print("INSUFFICIENT WEIGHTS")
        sys.exit()

    for i in range(c):
        weightage[i] /= s

    a = [0]*(c)

    for i in range(0, r):
        for j in range(0, c):
            a[j] = a[j]+(data[i][j]*data[i][j])

    for j in range(c):
        a[j] = m.sqrt(a[j])

    for i in range(r):
        for j in range(c):
            data[i][j] /= a[j]
            data[i][j] *= weightage[j]

    ideal_positive = np.amax(data, axis=0)
    ideal_negative = np.amin(data, axis=0)
    for i in range(len(impact)):
        if (impact[i] == '-'):
            temp = ideal_positive[i]
            ideal_positive[i] = ideal_negative[i]
            ideal_negative[i] = temp

    dist_pos = list()
    dist_neg = list()

    for i in range(r):
        s = 0
        for j in range(c):
            s += pow((data[i][j]-ideal_positive[j]), 2)

        dist_pos.append(float(pow(s, 0.5)))

    for i in range(r):
        s = 0
        for j in range(c):
            s += pow((data[i][j]-ideal_negative[j]), 2)

        dist_neg.append(float(pow(s, 0.5)))

    performance_score = dict()

    for i in range(r):
        performance_score[i+1] = dist_neg[i]/(dist_neg[i]+dist_pos[i])

    a = list(performance_score.values())
    b = sorted(list(performance_score.values()), reverse=True)

    rank = dict()

    for i in range(len(a)):
        rank[(b.index(a[i]) + 1)] = a[i]
        b[b.index(a[i])] = -b[b.index(a[i])]

    row = list(i+1 for i in range(len(b)))
    a = list(rank.values())
    rr = list(rank.keys())
    out = {'Row_NO': row, 'Performance_score': a, 'Rank': rr}
    output = pd.DataFrame(out)
    # print(output)
    dataset.drop(dataset.columns[0], axis=1)
    dataset.to_csv(resultFile, index=False)
    # print(dataset)
    df = pd.read_csv(resultFile)
    df["Topsis Score"] = a
    df.to_csv(resultFile, index=False)
    df["Rank"] = rr
    df.to_csv(resultFile, index=False)


if __name__ == "__main__":
    main()