File size: 1,403 Bytes
2e44da9 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 | # Forecasting Project Data Layout
This project is organized for research-first use.
## Folder map
- `Data/raw/minute/`
- Original minute OHLCV source files for NIFTY 50, NIFTY BANK, INDIA VIX, HDFCBANK, RELIANCE, WIPRO, and KOTAKBANK.
- `Data/processed/bars/`
- Clean OHLCV bars at `1m`, `5m`, `1h`, `4h`, and `1d`.
- `Data/processed/features/`
- Per-asset feature sets for each timeframe.
- `Data/processed/panels/`
- Multi-asset aligned panels plus the daily master panel.
- `Data/metadata/`
- `manifest.csv`, `feature_dictionary.csv`, `layout.json`, and this note.
## Feature design notes
- Price features include returns, log returns, candle body/range, ATR, RSI, MACD, Bollinger z-score, moving-average-relative levels, and next-bar targets.
- The supplied index minute files have zero volume throughout, so the pipeline suppresses volume features and adds `volume_all_zero` instead.
- Intraday bars are grouped inside each observed trading session to avoid overnight contamination.
## High-value outputs
- `Data/processed/panels/1m_market_panel.csv`
- `Data/processed/panels/5m_market_panel.csv`
- `Data/processed/panels/1h_market_panel.csv`
- `Data/processed/panels/4h_market_panel.csv`
- `Data/processed/panels/1d_market_panel.csv`
- `Data/processed/panels/daily_master_panel.csv`
## Rebuild
Run:
```powershell
python Code\scripts\data_preparation\build_research_data.py
```
|