How avinashrkrk/bitcoin-price-prediction works
Main Components
• Data Source & Preprocessing:
• Historical Bitcoin price data as the input. This component includes the raw data and the preprocessing steps (normalization, formatting for time-series).
• Data Analysis & Model Training:
• In the Jupyter Notebook (“cryptocurrency_data_analysis_and_prediction.ipynb”), show the steps from exploratory data analysis to training the LSTM neural network. Emphasize the use of libraries such as NumPy, Pandas, Matplotlib, and Scikit-learn in this phase.
• LSTM Model:
• Interactive Prediction (Streamlit App):
• Streamlit application (“app.py”) as the interactive UI that consumes the pre-trained LSTM model. This layer receives user input (e.g., date ranges or feature inputs) and outputs real-time price predictions along with visualizations.
• Visualizations & Output:
• A component for output/visualization (referencing the images that show the dashboard, performance metrics, and detailed predictions).
Relationships & Interactions
• Flow of Data:
• Interaction between Model & User Interface:
• Feedback loop (optional):
Architectural Patterns & Design Principles
• Separation of concerns:
• Data preprocessing and model training are handled in the Jupyter Notebook.
• The model serving and user interaction are segregated in the Streamlit app.
• Notes on the use of PyTorch for deep learning and Streamlit for rapid UI development.
• That the overall design supports extensibility – for instance, possible future integration with live data feeds or alternative models (e.g., GRU, Transformer).