IdeaCredIdeaCred

Tksrivastava/stable-customer-segmentation

55

Stable Customer Segmentation is an end-to-end ML pipeline that learns latent customer representations using autoencoders before applying clustering, enabling more stable and interpretable customer segmentation compared to traditional feature-space clustering approaches.

What's novel

Stable Customer Segmentation is an end-to-end ML pipeline that learns latent customer representations using autoencoders before applying clustering, enabling more stable and interpretable customer segmentation compared to traditional feature-space clustering approaches.

Code Analysis

6 files read · 2 rounds

A production-oriented customer segmentation system that learns stable latent representations via a deterministic autoencoder and clusters FMCG retail data using HDBSCAN in both raw and latent spaces.

Strengths

Strong separation of concerns with reusable core modules; substantive feature engineering (CV, entropy, YoY growth); deterministic design choices (seeds, L2 regularization) for production stability; clear pipeline orchestration.

Weaknesses

No test suite present; minimal input validation and error handling; hardcoded artifact paths reduce portability; Kaggle dependency adds friction for local testing.

Score Breakdown

Innovation
5 (25%)
Craft
45 (35%)
Traction
7 (15%)
Scope
60 (25%)

Signal breakdown

Innovation

Not Fork+1
Code Novelty+1
Concept Novelty+1

Craft

Ci-1
Tests-1
Polish+0
Releases+0
Has License+5
Code Quality+18
Readme Quality+15
Recent Activity+7
Structure Quality+5
Commit Consistency+2
Has Dependency Mgmt+5

Traction

Forks+0
Stars+6
Hn Points+0
Watchers+3
Early Traction+0
Devto Reactions+0
Community Contribs+0

Scope

Commits+7
Languages+5
Subsystems+10
Bloat Penalty+0
Completeness+7
Contributors+5
Authored Files+8
Readme Code Match+3
Architecture Depth+3
Implementation Depth+8

Evidence

Commits

21

Contributors

1

Files

16

Active weeks

5

TestsCI/CDREADMELicenseContributing

Repository

Language

Python

Stars

1

Forks

0

License

NOASSERTION