ADS Capstone Chronicles Revised
aim was to identify any multicollinearity issues. In the heatmaps below, the top right corner displays eleven boxes with a score of 1.00, colored red, indicating a strong positive correlation between these variables. Given the data context, the high, open, close, and average prices are all related to the price, which explains their high correlation with each other. The price change and percent change columns also display a strong positive correlation coefficient of 0.80. Moving down the heatmap towards the middle left, other highly correlated variables are observed, such as all the close ema
and close lag values. This is logical since these metrics are variations of each other.
Figure 2 and Figure 3 provide insight into which variables should be removed to address multicollinearity issues. Variables with a correlation coefficient above 0.70 or 0.75 should be considered for removal, and some metrics will be excluded due to significant null values. This plot includes two heatmaps, one for ADA-USD and one for BTC-USD, to show how similar they both are; however, the rest of the heatmaps almost all follow the same trend due to the metrics involved.
Figure 2 Heatmap for ADA-US
75
Made with FlippingBook - Online Brochure Maker