Topsis-Khushi-102017112 0.0.0

A python package to implement TOPSIS on a given dataset

TOPSIS-Python

Submitted By: Khushi Goel
Roll Number: 102017112
Batch: 3CS5

---

pypi: https://pypi.org/project/Topsis-Khushi-102017112
git: https://github.com/khushi-0001/Topsis-Khushi-102017112

---

Installation

pip install Topsis-Khushi-102017112

What is TOPSIS

TOPSIS or Technique for Order Preference by Similarity to Ideal Solution is a method of compensatory aggregation that compares a set of alternatives by identifying weights for each criterion, normalising scores for each criterion and calculating the geometric distance between each alternative and the ideal alternative, which is the best score in each criterion

How to use this package:

Topsis-Khushi-102017112 can be run as in the following example:

In Command Prompt

>> topsis input.csv "1,1,1,1,1" "+,+,-,+,+" output.csv

Process

First we create an evaluation matrix which consists of m alternatives and n criterias, with the intersection of each alternative and criteria. Then we move to the preprocessing phase. We then normalize the matrix using norm. Weighted normalised decision matrix is then calculated. We then determine the best and worst alternatives. After that, we calculate the euclidean distance between the target alternative and the worst condition. Finally, the similarity to the worst condition checked and the alternatives are ranked according to the final performance scores, awarding lower rank to higher performance score.

Sample dataset

Consider this sample.csv file

First column of file is removed by model before processing so follow the following format.

All other columns of file should not contain any categorical values.

Model	P1	P2	P3	P4	P5
M1	0.85	0.72	4.6	41.5	11.92
M2	0.66	0.44	6.6	49.4	14.28
M3	0.9	0.81	6.7	66.5	18.73
M4	0.8	0.64	6.9	69.7	19.51
M5	0.84	0.71	4.7	36.5	10.69
M6	0.91	0.83	3.6	42.3	11.91
M7	0.65	0.42	6.9	38.1	11.52
M8	0.71	0.5	3.5	60.9	16.4

weights vector = [ 1,2,1,2,1 ]

impacts vector = [ +,-,+,+,- ]

input:

topsis input.csv "1,2,1,2,1" "+,-,+,+,-" output.csv

output:

output.csv file will contain following data :

Model	P1	P2	P3	P4	P5	Topsis score	Rank
M1	0.85	0.72	4.6	41.5	11.92	0.3267076760116426	6
M2	0.66	0.44	6.6	49.4	14.28	0.6230956090525585	2
M3	0.9	0.81	6.7	66.5	18.73	0.5006083702087599	5
M4	0.8	0.64	6.9	69.7	19.51	0.6275096427934269	1
M5	0.84	0.71	4.7	36.5	10.69	0.3249142875298663	7
M6	0.91	0.83	3.6	42.3	11.91	0.2715902624653612	8
M7	0.65	0.42	6.9	38.1	11.52	0.5439263412940541	4
M8	0.71	0.5	3.5	60.9	16.4	0.6166791918077927	3

The decision matrix (a) should be constructed with each row representing a Model alternative, and each column representing a criterion like Accuracy, R², Root Mean Squared Error, Correlation, and many more.

Weights (w) is not already normalised will be normalised later in the code.

Information of benefit positive(+) or negative(-) impact criteria should be provided in I.

The rankings are stored in a csv file, with the 1st rank offering us the best decision, and last rank offering the worst decision making, according to TOPSIS method.

Debugging and Exception Handling

The program has several assert statements which raise errors with helpful description in the following cases:

• Wrong dimensions of decision matrix (not 2D), weights (not 1D)
• Length of weights and impacts don't match
• Weights or impacts don't match number of attributes
• For command line, number of arguments is less than 3 required
• File extension must be .csv

License

Copyright 2023 Khushi Goel
This repository is licensed under the MIT license.
See LICENSE for details. MIT