An Optimized Regularization Based Machine Learning Approach for Forecasting Crypto Prices

An Optimized Regularization Based Machine Learning Approach for Forecasting Crypto Prices

Rahul Hidau1, Prof. Ashish Tiwari2

Abstract— Machine Learning and Deep Learning Models are being extensively used at the backend of forecasting crypto prices. Many factors, including changes in regulation, public opinion on social media, investor actions, and overall global economic trends, contribute to the high degree of volatility in the cryptocurrency market. Predicting the future value of cryptocurrencies with any degree of accuracy has emerged as a top concern for academics, traders, and investors due to the inherent uncertainty of the market. Machine learning (ML) provides excellent methods for evaluating large volumes of real-time and historical data to forecast price changes in the future. ML models range from simple statistical methods to intricate deep learning architectures. The proposed work employs an optimized second order regularization based back propagation algorithm along with the data pre-processing using the discrete wavelet transform (DWT) for crypto price prediction. It has been shown that the proposed system attains lesser mean square percentage error compared to previously existing technique.

Keywords—Machine Learning, Neural Networks, Regularization, Time Series Analysis, Mean Absolute Percentage Error (MAPE).