https://github.com/w7negreiros/cryptoclustering

Uses K-Means clustering to group cryptocurrencies by their performance - Crypto Clustering Challenge - Unsupervised Learning - UofT Data Analytics - Bootcamp
https://github.com/w7negreiros/cryptoclustering

Last synced: 10 days ago
JSON representation

Uses K-Means clustering to group cryptocurrencies by their performance - Crypto Clustering Challenge - Unsupervised Learning - UofT Data Analytics - Bootcamp

• Host: GitHub
• URL: https://github.com/w7negreiros/cryptoclustering
• Owner: w7negreiros
• Created: 2024-07-17T14:30:09.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-07-17T15:33:48.000Z (7 months ago)
• Last Synced: 2024-12-04T02:13:31.240Z (2 months ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 215 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Clustering Cryptocurrencies - UofT Data Analytics - Bootcamp

In this project I apply the unsupervised learning technique of K-Means clustering to group cryptocurrencies by their performance in an effort to create profitable portfolio recommendations.

# Data Used

[crypto_market_data.csv](Resources/crypto_market_data.csv) - market data of different cryptocurrencies during different time periods

# Summary

I start by using the elbow curve method, using normalized data, to find the optimal k value for the K-Means model that will use all of the original features of the dataset.

![Resources/Crypto Visualizations/elbow_curve.png]()

Then, using the optimal k value I train and predict the K-Means model to generate 4 clusters of cryptocurrencies. The inertia of each cluster was significant enough to consider reducing the amount of features.

![image 2]()

To reduce the amount of features used, I applied Principal Component Analysis to create 3 primary clusters.

![image 3]()

I then used the PCA data to again calculate the optimal k value for the K-Means model.

![image 4]()

Finally, with the optimal k value for the PCA features, I plot the new clusters.

![image 5]()

# Technologies

This is a Python 3.7 project ran using a JupyterLab in a conda dev environment.

The following dependencies are used:

Jupyter - Running code 

Conda (4.13.0) - Dev environment    

Pandas (1.3.5) - Data analysis  

Matplotlib (3.5.1) - Data visualization 

Numpy (1.21.5) - Data calculations + Pandas support     

hvPlot (0.8.1) - Interactive Pandas plots   

scikit-learn (1.0.2) - KMeans clustering, data normalization, and PCA   

# Usage

The Jupyter notebook [crypto_investments.ipynb](Crypto_Clustering.ipynb) will provide all steps of the data collection, preparation, and analysis. Data visualizations are shown inline and accompanying analysis responses are provided.