Abstract
A semi-parametric approach is developed to model nonlinear time series. This approach is used on hourly electricity load data, it considers the effect of temperature combined with those of time and day type. The temperature effect is modelled by nonparametric smoothing. An ARIMA model is used to model the serial correlation. An iterative backfitting algorithm is proposed for estimation. The post sample forecasting performance and comparative results show good potential of this approach.
| Original language | American English |
|---|---|
| State | Published - 2002 |
| Event | ICSA Applied Statistics Symposium - Duration: Dec 15 2020 → … |
Conference
| Conference | ICSA Applied Statistics Symposium |
|---|---|
| Period | 12/15/20 → … |
Keywords
- Semi-parametric approach
- Nonlinear
- Time series
- Hourly electricity load data
- ARIMA
- Nonparametric smoothing
DC Disciplines
- Business