A Cryptocurrency Price Prediction using Linear Regression and LSTM

Cryptocurrency markets, particularly Bitcoin, exhibit high volatility and complex price dynamics that pose significant challenges for accurate forecasting. Reliable prediction of price trends is crucial for investors, traders, and financial analysts to make informed decisions. Traditional statistical models often fail to capture nonlinear patterns and temporal dependencies inherent in financial time-series data. This paper presents a comparative study of Bitcoin price prediction using two approaches: Linear Regression as a baseline model and Long Short-Term Memory (LSTM) networks as a deep learning model. We utilize historical Bitcoin price datasets, perform rigorous data preprocessing and normalization, and evaluate model performance using Mean Squared Error (MSE) and Root Mean Squared Error (RMSE). The experimental results demonstrate that while Linear Regression provides an interpretable baseline for trend identification, LSTM significantly enhances prediction accuracy by effectively capturing sequential dependencies and nonlinear market behaviour. The key contribution of this work is the systematic comparison of these two approaches with a structured pipeline covering data acquisition, feature engineering, model training, and performance evaluation. Furthermore, we analyze the limitations of current approaches and highlight potential improvements for building more robust real-time cryptocurrency prediction systems.