Installation

In order to run the notebooks in this repository, the following libraries have to be installed:

1. Pandas 0.24.2
2. Numpy 1.17.4
3. seaborn 0.9.0
4. scikit-learn 0.21.2
5. tensorflow 2.6.0
6. Flask 2.1.3
7. Plotly 5.11

Project Motivation

The purpose of this project is to cluster the input data. The input data consists of data pertaining to a particular commodity. Various clustering methods have been compared and suitable clustering approach has been selected based on the comparison results.

Technical Details

This project demonstrates:

1. Extensive exploratory data analysis
2. Use of tensorflow.keras to build an auteencoder
3. Use of evaluation metrics to compare clustering approaches
4. Use of CRISP-DM steps

File Descriptions

The repository consists of 2 main folders -- Data & Code
The Data folder has:

1. cmd_attributes_v3_upload.csv: This file has the pre-processed input data

The Code folder has:

1. CMD_Clustering_Steps1_2.ipynb: This is a jupyter notebook showing the first 2 steps of CRISP-DM (Business & data understanding)
2. CMD_Clustering_Steps3_5.ipynb: This is a jupyter notebook showing steps 3, 4 and 5 of CRISP-DM (Data preparation, modeling & evaluation)
3. cmd_data_output.csv: csv file with cluster information and is used in the python script
4. CMD_Clustering_Deploy_Step6.py: This is a python script for a web-app showing details about the clusters
5. index.html & chart_4.html: HTML templates used in the webapp
6. Clustering_Project_v4.pdf: Document describing the overall approach for clustering

Steps for running the python script:

1. cd SKU_Clusters ## go to the location of the repository
   
2. python ./Code/CMD_Clustering_Deploy_Step6.py

Acknowledgements

Thanks to Python open source community for creating valuable libraries used in this project.