amyylin1 / cryptocurrencies

Cryptocurrencies

Introduction

This analysis uses unsupervised machine learning and principal components analysis (PCA) to decide which cryptocurrency to invest in.

Results

PCA

The analysis is done with three principal components. However, all three principal components explained the overall variances poorly. The 1st principal component accounts for 2.7% of the overall variability; the 2nd principal component accounts for 2.1% of the overall variability; and the 3rd principal component accounts for 2.0% of the overall variability (see figure below).

K-Means

Based on the Elbow curve below, the best value for k is 4, meaning that 4 clusters are ideal.

3D-scatter with clusters

3D-scatter plot shows there are four distinct clusters.