Bitcoin_Price_Prediction_Presentation.pptx
- 1. Bitcoin Price Prediction Using ARIMA Model Project Presentation
- 2. Abstract • Bitcoin is a cryptocurrency built on blockchain technology. Predicting its price is challenging due to volatility. This project uses ARIMA time- series model with historical data (2014– Present) to forecast Bitcoin prices.
- 3. Introduction • • Bitcoin is a decentralized digital currency. • • Price influenced by market demand, regulations, investor sentiment. • • High volatility makes prediction challenging. • • Machine learning and time-series analysis help improve forecasts.
- 4. Literature Review • • McNally et al. (2018): LSTM networks for Bitcoin prediction. • • Kristjanpoller & Minutolo (2018): GARCH model for volatility. • • Mallqui & Fernandes (2019): Random Forest, Naïve Bayes models. • • ARIMA remains effective for trend-based forecasting.
- 5. Objectives • • Collect and preprocess Bitcoin price data (2014–Present). • • Analyze trends using time-series tools. • • Develop predictive model with ARIMA. • • Evaluate model accuracy. • • Provide insights into cryptocurrency forecasting.
- 6. Methodology • • Data Source: Coindesk BTC Price Index. • • Data preprocessing: Handle missing values, stationarity tests. • • Model: Implement ARIMA using Python statsmodels. • • Evaluation: MSE, RMSE, MAE metrics. • • Tools: Python, Pandas, NumPy, Matplotlib, Statsmodels.
- 7. Hypothesis • • Null Hypothesis (H₀): ARIMA does not significantly predict future Bitcoin prices better than random chance. • • Alternative Hypothesis (H₁): ARIMA significantly predicts Bitcoin prices better than random chance.
- 8. Conclusion • • Predicting Bitcoin price is complex due to volatility. • • ARIMA captures historical trends and provides reliable forecasts. • • Model is limited in handling sudden market shocks. • • Can be extended with advanced ML models (LSTM, hybrid models).
- 9. Future Scope • • Explore deep learning models (LSTM, GRU). • • Integrate real-time market sentiment analysis. • • Enhance prediction accuracy with hybrid models. • • Deploy as a web application using Streamlit & AWS.