topsis-sachingoyal-102303557 1.0.0

TOPSIS decision method Python package

TOPSIS Implementation (Sachin Goyal)

Simple TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) implementation in Python.

Project structure

• data.csv (example input)
• setup.py (packaging / install - optional)
• topsis/ (package)
  • init.py
  • topsis.py (main implementation and CLI)

Summary

• Reads a CSV where the first column is an identifier (name) and the remaining columns are numeric criteria.
• Computes TOPSIS scores and ranks using provided weights and impacts.
• Writes an output CSV that includes the original data plus Topsis Score and Rank columns.

Prerequisites

• Python 3.6+
• pandas and numpy

Quick install (no virtualenv):

Open a terminal and run:

python -m pip install pandas numpy

Or create a virtual environment first, then install the same packages.

Usage

There are two simple ways to run the script.

1. Run as a module (from project root):

   python -m topsis.topsis "" ""

2. Run the script file directly:

   python topsis\topsis.py "" ""

Arguments

• : path to the input CSV file.
• : comma-separated numeric weights for each criterion (e.g. "1,1,1,1").
• : comma-separated + or - for each criterion (e.g. "+,+,-, +").
• : path for the output CSV.

Input CSV format

• Must have at least 3 columns (first column: identifier; at least two criteria columns).

• All columns from the 2nd to last must contain numeric values.

• Example:

  Name,Cost,Performance,Reliability A,250,8,9 B,200,7,8

Example

python -m topsis.topsis data.csv "1,1,1" "+,+,-" result.csv

This will produce result.csv with two extra columns:

• Topsis Score (higher is better)
• Rank (1 = best)

Common errors and messages

• "Input file not found": the specified input file path doesn't exist.
• "Unable to read input file": input file not a valid CSV.
• "Input file must contain at least 3 columns": CSV missing required columns.
• "All columns from 2nd to last must contain numeric values": non-numeric values in criteria columns.
• "Number of weights must be equal to number of criteria": mismatch between provided weights and criteria count.
• "Number of impacts must be equal to number of criteria": mismatch between provided impacts and criteria count.
• "Impacts must be either + or -": impacts should be "+" or "-" only.
• "Weights must be numeric": weights should be numbers.

Notes

• Weights are normalized implicitly by the algorithm's normalization step.
• The implementation uses Euclidean normalization.

Author

• Sachin Goyal (student project)

License

• Use or adapt as needed for learning or coursework.