Cryptocurrencies

Overview:
Using Unsupervised Machine learning to predict Crytocurrency investment trends for new and less prominent currencies. Unsupervised learning is the best way forward for this dataset because looking for any groupings, trends, or other information help identify cryptocurrencies to investment firms.

Preprocessing the Data for PCA

Cleaned dataframe, dropped null and n/a values, dropped "Is Trading" column

scaled data to run using PCA

Reducing Data Dimensions Using PCA

used PCA to reduce dimensions to three principal components

Clustering Cryptocurrencies Using K-means

Using this image, chose a k=5 where the bend of the 'elbow' occurs.

Visualizing Cryptocurrencies Results

PCA data and clusters of classes predicted in the cryptocurrencies

Looking at the outliers in this scatterplot would make a difference if there was a reason to drop them from the data set and take a closer look at the rest of the cluster in more detail.

knittelc / Cryptocurrencies