Forecast Variance Analysis

Forecast variance analysis compares the actual historical values to the forecasted values so that neural network definitions that are not performing well can be identified and retrained with the assumption that future estimates will improve. The analysis process runs at a time that is after the forecasted time interval elapses and the actual output values become known. This allows the variance between estimated and actual output values to be determined. If the variance is too high, retrain the neural network with an improved set of historical input and output values.

Variance analysis processing can add retraining tasks to the training priority queue for neural network definitions whose forecasts are not within an acceptable range of the actual energy load.

The forecast variance analysis process can be triggered on demand for all qualified neural network definitions by a direct user command or scheduled through the MSS.

Configuration

• Variance Analysis commands are configured on the UIS Commands page of the ELF Editor. See Configuring UIS Commands.

• A default system Variance Threshold can be configured on the Systems Settings page of the ELF Editor. See Configuring System Settings.

• This Variance Threshold value can be overridden at the neural network template and definition level on the Tuning Parameters dialog box. See Configuring Tuning Parameters.

• The Variance status of the neural network training process is displayed on the Neural Network Definition Settings dialog box. See Configuring Neural Network Definition Settings.

• The Variance Analysis log can be viewed on the Device page. See Configuring the ELF Device.