Comparative Analysis of LSTM and XGBoost Models for Short-Term Bitcoin Price Prediction

Abstract

As the cryptocurrency continues to evolve, accurate prediction of cryptocurrency prices has become a vital area of interest. Cryptocurrency markets, known by their rapid fluctuations, present challenges for predictive modeling. Accurate prediction of cryptocurrency prices is challenging yet crucial task for investors and stakeholders. In this objective, LSTM, a recurrent neural network architecture well known for its sequence modeling capabilities, is evaluated in parallel against XGBoost, a gradient boosting algorithm renowned for its proficiency in structured data analysis and predictive accuracy. Evaluation metrics guide the analysis, enables assessing of models’ performance. Metrics reveal the strengths and limitations of LSTM and XGBoost models, enriching understanding of their applicability. In this study, we present a comparative analysis of two popular machine learning techniques, Long Short-Term Memory (LSTM) and Extreme Gradient Boosting (XGBoost) and determine the best model among them and also Bitcoin prices for the next 30 days is forecasted using XGBoost