Topsis-Anvvi-102317133 1.0.0

TOPSIS multi-criteria decision making — CLI tool with sensitivity dashboard

Topsis-Anvvi-102317133

A Python package implementing the TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) multi-criteria decision analysis method. Supports CSV and Excel input, automatic ordinal encoding, null handling, and an optional interactive sensitivity analysis dashboard.

---

Table of Contents

• Installation
• Method Overview
• Command-Line Usage
• Input File Format
• Ordinal Encoding
• Null Handling
• Output Format
• Sensitivity Analysis Dashboard
• Programmatic API
• Requirements
• License

---

Installation

pip install Topsis-Anvvi-102317133

---

Method Overview

TOPSIS ranks a set of alternatives by measuring how close each one is to an ideal best solution and how far it is from an ideal worst solution. The algorithm proceeds as follows:

1. Normalise the decision matrix using Euclidean (column-wise) normalisation
2. Apply weights to produce a weighted normalised matrix
3. Determine the ideal best and ideal worst vectors, respecting each criterion's impact direction (+ beneficial, - cost)
4. Compute Euclidean distances from each alternative to both ideal vectors
5. Calculate the TOPSIS score — a value in [0, 1] where higher is better:

Score(i) = dist_worst(i) / (dist_best(i) + dist_worst(i))

6. Rank alternatives by score in descending order

---

Command-Line Usage

topsis <InputFile> <Weights> <Impacts> <ResultFile> [--sensitivity]

Arguments

Argument	Type	Description
InputFile	str	Path to input .csv or .xlsx file
Weights	str	Comma-separated positive numbers e.g. "1,1,2,1"
Impacts	str	Comma-separated + or - signs e.g. "+,+,-,+"
ResultFile	str	Path for output .csv or .xlsx file
--sensitivity	flag	Optional — generates an interactive HTML dashboard

Examples

# Basic usage
topsis data.csv "1,1,2,1" "+,+,-,+" result.csv

# With sensitivity dashboard
topsis data.csv "1,1,2,1" "+,+,-,+" result.csv --sensitivity

# Excel input and output
topsis data.xlsx "1,1,2,1" "+,+,-,+" result.xlsx --sensitivity

---

Input File Format

The input file must follow this structure:

• Column 1 — Alternative names (any string label)
• Columns 2 onward — Criteria values (numeric, or ordinal text — see below)
• The number of criteria columns must equal the number of weights and impacts provided

Sample Input (data.csv)

Model	Cost ($)	Storage (GB)	Camera (MP)	Battery	Rating
M1	250	16	12	5000	good
M2	200	32	13	3000	poor
M3	300	32	12	3500	good
M4	275	32	17	2500	average
M5	225	16	13	4800	excellent

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

Cost has impact - (lower is better). All others have impact + (higher is better). The text column Rating is automatically encoded.

---

Ordinal Encoding

Non-numeric text columns are automatically detected and encoded to integers if they match a known ordinal scale:

Level	Scale Values	Encoded As
2	no, yes	0, 1
2	false, true	0, 1
3	low, medium, high	1, 2, 3
3	bad, moderate, good	1, 2, 3
3	poor, average, good	1, 2, 3
3	weak, moderate, strong	1, 2, 3
3	small, medium, large	1, 2, 3
4	poor, average, good, excellent	1, 2, 3, 4
4	low, medium, high, very high	1, 2, 3, 4
5	very bad, bad, moderate, good, very good	1, 2, 3, 4, 5
5	very low, low, medium, high, very high	1, 2, 3, 4, 5
5	very poor, poor, average, good, excellent	1, 2, 3, 4, 5

Matching is case-insensitive and trims whitespace. If a column contains text that doesn't match any known scale, the program exits with a descriptive error.

---

Null Handling

Missing values are handled automatically before computation:

• Rows where all criteria are null — dropped entirely, with a warning printed
• Numeric columns with nulls — filled with the column mean
• Text columns with nulls — filled with the column mode

All substitutions are reported to stdout.

---

Output Format

The result file contains all original columns plus two appended columns:

Model	Cost ($)	...	Topsis Score	Rank
M5	225	...	0.7623	1
M1	250	...	0.6891	2
M3	300	...	0.5340	3
M4	275	...	0.4102	4
M2	200	...	0.2987	5

• Topsis Score — float in [0, 1]; higher = better
• Rank — integer; 1 = best; ties share the same rank (dense ranking)

---

Sensitivity Analysis Dashboard

When --sensitivity is passed, a self-contained HTML file is generated alongside the result (e.g. result_sensitivity.html). Open it in any browser — no server required.

Features

• Live weight sliders — adjust each criterion's weight in real time; rankings update instantly
• Score bars — visual comparison of TOPSIS scores across all alternatives
• Rank delta badges — shows how many positions each alternative moved from the original ranking
• Sweep analysis chart — plots how every alternative's rank changes as one criterion's weight is swept from 0 → 5, while all others are held fixed
• Natural-language insight — auto-generated summary of the most significant rank change observed

Interpreting the sweep chart

A flat horizontal line means an alternative's rank is robust to changes in that criterion. Crossing lines indicate rank reversals — points where two alternatives swap positions as the weight shifts.

---

Programmatic API

TOPSIS can also be called directly from Python:

from Topsis_Anvvi_102317133.Topsiss import topsis

result_df = topsis(
    input_file  = "data.csv",
    weights_str = "1,1,2,1",
    impacts_str = "+,+,-,+",
    result_file = "result.csv",
    sensitivity = True      # False to skip dashboard
)

print(result_df[["Topsis Score", "Rank"]])

The function returns the result DataFrame directly, in addition to writing it to disk.

---

Requirements

Package	Minimum Version
Python	3.7+
pandas	1.3.0
numpy	1.21.0
openpyxl	3.0.0

---

License

MIT License — free to use, modify, and distribute.