Project Information
- Category: Data Analytics (Time Series Forecasting)
- Project URL: GitHub Repository
Objective
To measure how daily temperatures affect electricity demand. This project also predicts how accurately temperature could be used to forecast the demand of electricity in mid-term (8 weeks/2 months).
- Techniques:
- Periodogram of residuals
- Auto-correlation and Spectral Density
- Cross-correlation and Coherence
- F-Test / Analysis of Variance (ANOVA) Test
- A/B Testing
- AIC/BIC metric
- Models Used:
- Linear Difference Equations with different orders (lags) and cyclic components
- Auto Regressive Moving Average (ARIMA/ARMAX) with different orders (lags) and cyclic components