Project Summary

Problem 1: Clustering

Section 1

• Used historical data for top 4 market coins over a one-year period.
• Implemented K-means clustering with market cap and volume features.
• Visualized clusters on a scatter plot.

Section 2

• Explored optimal k-values using K-means algorithm (k from 1 to 10).
• Emphasized importance of choosing k and encouraged innovative methods.

Section 3

• Applied DBScan method for clustering based on market cap and volume.
• Adjusted hyperparameters for 5 meaningful clusters.
• Visualized clusters and explained hyperparameter impact.

Problem 2: Hierarchical Clustering

Section 1

• Conducted hierarchical clustering with market cap and volume.
• Visualized results in a dendrogram.
• Identified clusters for 2-cluster division and provided analysis.

Section 2

• Incorporated ProofType attribute for three-dimensional hierarchical clustering.
• Compared and interpreted clustering results with previous state.

Section 3

Introduced relevant attributes for more meaningful clustering.

• Encouraged inclusion of any pertinent attributes for improved interpretation. -

Problem 3: Prediction

• Focused on predicting Monero's closing price increase or decrease for the next day.
• Utilized yfinance library for cryptocurrency data extraction.
• Allowed use of machine learning algorithms taught in bootcamp classes.
• Emphasized target variable labeling based on final price comparison.
• Suggested potential features for dataset, ensuring no target variable leakage.

The project aimed to enhance understanding of clustering algorithms, hierarchical clustering, and predictive modeling in the context of cryptocurrency data, while encouraging creative problem-solving and exploration of diverse features for improved predictions.