Model updated 5th February 2024

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%. (see validation results table below comparing the previous models with the new models).

Training period

Both PD and 7day price forecast models were trained on data between 01-01-2021 and 15-01-2024.

The market suspension period and a period after the Callide explosion were removed from the training data.

Validation set

The model was validated by predicting the price between 16-01-2024 and 30-01-2024

New features (input data) exceptions

• The feature set is the same as the previous model. The main benefit of the new model is that more recent market data are consumed in the training process. The training does not weight the most recent market data.

  • A number of experiments were conducted by changing the feature list and some hyper parameters however this didn’t not achieve a significant benefit to the model tests. Hence for consistency we have not made changes to the model other than extending the training period to the most recent market data. Unlike the August model update included significant changes to both the feature set and the hyper parameters.

• Validation set is from the 1st of November to the 7th of November. Oddly enough the impact of the Melbourne Cup did not have a significant impact on the performance metrics.

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 volatile.

Average error is equal to Predication minus real prices.

All error metrics are in $/MWhr.

PD Results

7 Day Results

Model updated 9th November 2023

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 between 10% and 15% (with the exception of SA which was 50%). (see validation results table below comparing the previous models with the new models).

Training period

• Both PD and 7day price forecast models were trained on data between 01-11-2020 and 01-11-2023.

  • The market suspension period and a period after the Callide explosion were removed from the training data.

New features (input data) exceptions

• The feature set is the same as the previous model. The main benefit of the new model is that more recent market data are consumed in the training process. The training does not weight the most recent market data.

  • A number of experiments were conducted by changing the feature list and some hyper parameters however this didn’t not achieve a significant benefit to the model tests. Hence for consistency we have not made changes to the model other than extending the training period to the most recent market data. Unlike the August model update included significant changes to both the feature set and the hyper parameters.

• Validation set is from the 1st of November to the 7th of November. Oddly enough the impact of the Melbourne Cup did not have a significant impact on the performance metrics.

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).

Average error is equal to Predication minus real prices.

All error metrics are in $/MWhr.

PD Results

7 Day Results

8th August 2023 release

Summary

The retraining process resulted in a new models, one for PD and one for 7day price forecast. After considerable experimentation we retrained the price forecasting models with a considerable increase in accuracy (see validation results table below comparing the previous models with the new models).

Training

• Both PD and 7day price forecast models were trained on data between 01-10-2020 and 20-08-2023.

  • The market suspension period and a period after the Callide explosion were removed from the training data.

New features (input data) exceptions

• A new “recent prices feature” was added to give the model some context for the current pricing environment. This feature is the median price for a window around the forecast time for the most recent three days. For example, for a forecast occurring in 2 days at 11am, the model generated the median value for prices between 10am-12pm for the three days before the forecast was run.

• A number of additional experiments were conducted on the 7 day forecast model, with the major modifications including:

  • The 50th percent probability of demand features were augmented with 10th percent probability of demand data

  • The new recent prices features mentioned previously

  • A new time of day feature

  • A new weekday/weekend feature.

• e Cup did not have a significant impact on the performance metrics.

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).

All error metrics are in $/MWhr.

PD Results

7 Day Results

Previous release history

6th June 2023

Summary

The current model was trained using data up until 30th March 2023. Retraining the model included a months data after the retirement of Liddell. We also conducted many experiments by changing the feature set (input variables) as well as boosting (weighting) more recent training periods.

Much to our surprise the performance of all retrained and new models demonstrated that the original model either performed better or more or less equal to all new models when measured against the validation set (which was a week of recent data not used in the training set). As a result we felt it best to keep the current model. Therefore we did not change the model as scheduled for the 6th of June.

Why retraining did not result in a better model (for 6th June)

There are a number of reasons why the original model performed better than all other retrained models however we believe that the this was due to the volatile market dynamics post Liddell retirement and that the market behaviour has since settled. Models that included the weeks after the retirement of Liddell are less representative of current market dynamics and now overstate forecast prices relative to actual market outcomes.