Model updated
...
28th January 2025
Summary
The retraining process resulted in two new models, one for PD and one for 7day price forecast. After experimentation we retrained the price forecasting models using existing features and hyper parameter settings. The increase in accuracy (using Absolute Mean Error, AME) was approximately 5%main benefit of the retrained model is that it includes the most recent market outcomes which will therefore be factored into the model. (see validation results table below comparing the previous models with the new models).
...
Both PD and 7day price forecast models were trained on data between 01-0701-2021 2022 and 0801-0724-20242025.
The market suspension period and a period after the Callide explosion were removed from the training data.
Validation set
The This model was validated by predicting the price between 09-07-2024 and 21-07-2024not validated, rather we felt that it was better to train the model up to the most recent time period. Earlier models have maintained the feature set and hyper parameter settings.
New features (input data) exceptions
...
Predispatch model | p7day model | |
|---|---|---|
Model_id | 2024050320250121_090016160406_CB_2 | 2024050320250121_085256160406_CB_7D_2 |
Latest release date (after retraining) | 26th July 202428th January 2025 | 28th January 2025 |
Previous model release date | 26th July 12th November 202426th July | 12th November 2024 |
Machine Learning Algorithm | Ensemble predictor: gradient boosting | Ensemble predictor: gradient boosting |
Training Period | 1st July 2021 to 8th July 20241st July 2021 to 8th July 2024January 2022 to 24th January 2025 | 1st January 2022 to 24th January 2025 |
Validation Period | N/A | N/A |
Hyper-parameter type | A | A |
Hyper-parameter notes | No weighting across time periods | No weighting across time periods |
Target | Dispatchprice (30min resolution) | Dispatchprice (30min resolution) |
Input (and training) variables | PREDISPATCHPRICE rrp DISPATCHPRICE last 3 day average actual RRP for same time PREDISPATCHREGIONSUM (intervention = 0) hours_out time_of_day day_type (weekday_or_weekend) dispatchableload netinterchange (availablegeneration - uigf) (dispatchablegeneration - scheduled_clearedmw) ss_solar_clearedmw ss_solar_uigf ss_wind_clearedmw ss_wind_uigf totaldemand totalintermittentgeneration | STPASA_REGIONSOLUTION hours_out time_of_day day_type (weekday_or_weekend) aggregatepasaavailability aggregatescheduledload netinterchangeunderscarcity aggregatecapacityavailable ss_solar_cleared ss_wind_uigf ss_wind_cleared ss_solar_uigf demand10 demand50 totalintermittentgeneration demand_and_nonschedgen |
Validation Results
These are the results of the best model generated through our experimentation process and compares the results with the previous model across our validation set (our test period).
Note that errors are considerably higher because the validation set was unusually volatilePrevious experimentation demonstrates that the current model features and settings provides a pretty good model across most market conditions. Therefore a validation set was not used rather the new model was trained right up to the latest market outcomes.
Average error is equal to Predication minus real prices.
...