This repository delivers an end-to-end econometric time series pipeline to model and forecast Turkey’s national electricity generation and consumption metrics (TEİAŞ). By decoupling complex multi-seasonal grid behaviors, the framework evaluates structural trends, applies comprehensive stationarity diagnostics, and deploys tuned SARIMAX models to ensure grid load optimization and energy security analytics.
- Goal: Model structural electricity consumption patterns, isolate temporal seasonality, and execute statistical out-of-sample projections with calibrated confidence boundaries.
- Expert Consensus (2026): High-frequency utility and grid data exhibit heavily volatile seasonal signatures. The modern analytical standard leverages robust STL (Seasonal and Trend Decomposition using Loess) to separate deterministic trend components from stochastic residuals, combining it with regularized SARIMAX wrappers to account for calendar effects and exogenous shocks before deploying machine learning layers.
- Operational Cleaning: Handles outlier detection and variance stabilization across cross-sectional generation metrics.
- Robust STL Decomposition: Extracts underlying structural trends and isolates distinct monthly/annual periodicities using localized regression modeling.
- Statistical Rigor: Runs simultaneous Augmented Dickey-Fuller (ADF) and KPSS diagnostics to guarantee mathematical transformation stationarity.
- Residual Verification: Validates model robustness using Ljung-Box and Jarque-Bera matrices to confirm residuals follow a strict non-correlated white noise distribution.
The repository strictly adheres to clean, production-ready modular machine learning layouts:
Time-Series-Project/
├── notebooks/ # Main interactive econometric research environment (teias_report.ipynb)
├── reports/
│ ├── html/ # Production-grade high-fidelity analytical report html
│ └── figures/ # Generated diagnostic assets (ACF, PACF, STL, Forecast Profiles)
├── presentation/ # Master executive slide deck (TEIAS_Final_Sunum.pptx)
├── .gitignore # Strict exclusion matrix for local JREs and datasets
└── README.md # Global architectural documentation