laurenemilyto / cryptocurrency-clusters

Cryptocurrency classification system using dimensionality reduction with PCA & t-SNE and cluster analysis with K-Means

Cryptocurrency Clusters

Background

• Created a report that includes what cryptocurrencies are on the trading market. Determined whether they can be grouped to create a classification system for this new investment.

• Processed data to fit t-SNE & k-Means.

Instructions

Data Preparation

• Read crypto_data.csv into Pandas. The dataset was obtained from CryptoCompare.

• Discarded all cryptocurrencies that are not being traded.

• Removed all rows that have at least one null value.

• Filtered for cryptocurrencies that have been mined.

• Converted features with text values into numerical data.

• Standardized dataset so that columns that contain larger values do not unduly influence the outcome.

Dimensionality Reduction

• Performed dimensionality reduction with PCA. Preserved 90% of the explained variance in dimensionality reduction.

• Reduced the dataset dimensions with t-SNE.

Cluster Analysis with k-Means

• Created an elbow plot to identify the best number of clusters.
• Used Silhouette Score to determine optimum number of clusters.
• Used heat map to identify highly correlated variables.

Recommendation

Based on my findings, cryptocurrencies can be clustered together into 4 distinct groups. Cluster results are grouped into csv files in the Clusters folder for further analysis.

Given more time, I would analyze these clusters for segmentation trends, looking at the global average of total coins mined and total coin supply, data compared to class average for each column.