How michaeldrah/pinn_ewc_dissertation works
This repository is a financial time-series modeling research project, organized around notebooks and persisted datasets, models, and evaluation outputs. Its visible stages include data acquisition and exploration, preprocessing and feature work, physics-informed modeling, continual-learning experiments, backtesting, and evaluation. The tree identifies these components but no README or source excerpts document calls or data transfers, so the graph does not assert relationships between them.
Components 22
Research notebooks
- Data download
notebooks/02_data_download.ipynb - Raw data exploration
notebooks/03_raw_data_exploration.ipynb - Dissertation EDA
notebooks/03a_dissertation_eda.ipynb - Preprocessing
notebooks/04_preprocessing.ipynb - Feature engineering
notebooks/05_feature_engineering.ipynb - Feature selection
notebooks/05a_feature_selection.ipynb - PINN modeling
notebooks/06_pinn_model.ipynb - Physics-informed loss
notebooks/06a – Physics-Informed Loss.ipynb - EWC module
notebooks/07_ewc_module.ipynb - Adaptive finance EWC
notebooks/07a_adaptive_finance_informed_ewc.ipynb - Backtesting
notebooks/08_backtesting.ipynb - Evaluation
notebooks/09_evaluation.ipynb
Market data
- Raw market data
data/raw - Preprocessed data
data/processed - Feature datasets
data/features - Model-ready datasets
data/model_ready
Model artifacts
- Trained model artifacts
models
Evaluation outputs
- Predictions
outputs/predictions - Metrics
outputs/metrics - Backtest results
outputs/backtesting - Research tables
tables
Other components
- Researcher