ADS Capstone Chronicles Revised

potential impact on cryptocurrency price prediction.

in capturing the intricate patterns within the data. The meticulous handling of missing values and multicollinearity further contributed to the model ’ s accuracy and robustness. 9.0.2 Forecasting Challenges The study also highlighted forecasting challenges in the highly volatile cryptocurrency market. While the model performed well overall, its underperformance in predicting BTC-USD emphasizes the need for specialized approaches to handle Bitcoin ’ s unique behavior. Additionally, the limitations of daily data granularity were evident, suggesting that higher frequency data could improve model accuracy. 10. Recommendations and Next Steps Based on the findings, several recommendations and future steps are proposed: 10.0.1 Use Higher Frequency Data Future studies should consider using higher frequency data, such as hourly or minute-by minute prices, to capture the rapid fluctuations in cryptocurrency markets more effectively. This approach promises to improve the accuracy of cryptocurrency price prediction models, offering a bright future for the field and instilling optimism in the audience. 10.0.2 Integrate Sentiment Analysis Incorporating sentiment analysis from social media platforms like Reddit and Twitter can provide additional insights into market movements. This approach could help identify bullish or volatile periods, significantly enhancing the predictive accuracy of cryptocurrency price prediction models. It enlightens the audience about its potential.

8.0.6 Conclusion

In conclusion, while this study has made significant strides in cryptocurrency price prediction using Ridge Regression, further research and model refinement remain necessary. By addressing the limitations and incorporating additional data sources, the goal of achieving highly accurate and reliable forecasting models for the cryptocurrency market can be met. These models could potentially revolutionize the understanding and prediction of price movements in the cryptocurrency market, providing valuable insights for investors and market analysts. 9 Conclusion The primary conclusion from this study is that Ridge Regression effectively forecasts cryptocurrency price movements. Despite initial expectations favoring more complex models like XGBoost and Random Forest, Ridge Regression emerged as the most consistent performer across all metrics. Its ability to handle multicollinearity and maintain robustness in volatile market conditions makes it particularly suited for cryptocurrency price prediction. This model demonstrated high R-squared values and low error rates, underscoring its reliability. 9.0.1 Feature Engineering Another significant finding is the importance of feature engineering in enhancing model performance. This includes ‘ lag features, ’ which are past values of the target variable used as predictors, and ‘ rolling statistics, ’ which are statistical measures calculated over a rolling data window. These techniques played a crucial role

89