Short-term load forecasting
Forecast 24hr horizon electrical load at 15-min intervals.
1) Data Preprocessing
Filled in missing values for temperature:
Time-aware linear interpolation
Fixed erroneous load data:
Replaced zero values with day-of-week / month mean for specific interval.
Divided data in a) training-validation and, b) test set
2) Model Formulation
Explored relationship of time - temperature - load
Random Forest -- Perform good with non-linear data, easy tuning (given time constraint) and implementation, scalable for this application. Working hypothesis about the model: Multiple independent models perform better than an aggregate since long-term exposure would introduce trend bias.
Temporal features (ie. previous day interval, morning / afternoon peak, min / max temperature)
3) Model tuning
Built methodology to cross-validate
See notebooks/02 for details.
4) Test model
Takes forecaster class and test set as parameter for out-of-sample validation. Used Mean Absolute Percentage Error
Sample out-of-sample prediction: MAPE = 1.8 %
5) Ongoing thoughts
Still high variance, cross-validate features, tree parameters, increase robustness of special days.
Temperature should be estimated through 2 variable interpolation of load-temperature.
Explain static load on weekdays.