Vignette-Clustering-Methods

Vignette on comparing three clustering methods (k means, spectral, agglomerative); created as a class project for PSTAT197A in Fall 2024.

Contributors

• Kasturi Sharma
• Nikhil Kuniyil
• Daniel Yan
• Johnson Sy Leung
• Lucas Joseph

Vignette Abstract

For this project, we are showcasing how clustering methods can be applied to various datasets. For our specific vignette, we chose the user behavior dataset that lists a person's phone model, operating system, app usage, battery life, etc. Our objective for this vignette is compare three different clustering methods -- k means, spectral, and agglomerative -- to group users based on similarities on their usage patterns. We will go through how to implement each of three methods by using this data set.

Repository Contents

• data contains

  • user_behavior_dataset.csv raw data

• scripts contains

  • knn_script.py script used to generate the processed data

  • spectral_script.py concise version of in-class codes used to replicate published analysis

  • agglomerative_script.py concise version of in-class codes used to replicate published analysis

• results contains a report template primary-vignette.ipynb and primary-vignette.html

• img contains graphs from the various scripts

Reference List

For further information on clustering methods implemented on Python, here are some links that may help!

1. https://scikit-learn.org/1.5/modules/clustering.html
2. https://www.geeksforgeeks.org/ml-spectral-clustering/
3. https://medium.com/@khalidassalafy/agglomerative-hierarchical-clustering-a-study-and-implementation-in-python-fddfdb6a7a64
4. https://medium.com/@amit25173/advanced-techniques-in-k-means-clustering-bd8cfa27ebc1
5. https://medium.com/swlh/k-means-clustering-on-high-dimensional-data-d2151e1a4240