ADS Capstone Chronicles Revised

19

from July 1st to July 8th demonstrates a downward trend from Monday to Thursday, followed by a significant increase in the average percent change from Thursday to Saturday and a sharp drop on Sunday. This pattern, observed across all analyzed

cryptocurrencies, suggests Saturday may be an optimal time for traders to sell coins, as prices tend to peak. Conversely, the period from Sunday to Wednesday might be more favorable for purchasing coins when prices are generally lower.

Figure 9 Average Daily Percentage Change by Day of the Week

7.0.6 Forecasted vs Actual Values In the final page of the Streamlit application, the forecasted values from the Ridge Regression model are compared to unseen data that was pulled from the API and not used during modeling. This visualization is intended to demonstrate how well or poorly the model performed on unseen data in a way accessible to individuals without a data science background.

with the forecasted values. The model could have forecast this specific coin more accurately. Upon further analysis, comparing actual prices to forecasted prices on unseen data, it can be concluded that the model is most likely overfitted. While the model performs well on some coins, it could perform better on Bitcoin. The indication that the model is overfitted arises from the observation that the forecasted values for each coin exhibit a similar trend with slight variation,

In Figure 10, the actual Bitcoin prices for the week of July 12th to the 18th are overlaid

86