Backtesting

When you create a forecast model, a backtest is automatically generated based on a subset of historical data (actuals). You can turn on the Write backtest results to Anaplan switch so the backtest results go into your forecast results module.

Further details:

• The data withheld is equal to the length of the forecast horizon. 
• The trained forecast model automatically produces predictions for the period represented by the withheld data. 
• Backtesting supports the comparison of forecast against actuals. The accuracy metrics derive from this comparison.

You have 12 months worth of historical data and a three month forecast horizon. 

• The model training process uses the first nine months of historical data. It also  predicts the remaining three months (this block of time is held back as a comparison sample). 
• Forecaster compares the forecast of the last three months with the actuals. This comparison results in advanced accuracy metrics that estimate the performance of the forecast model.

You can import the backtest data to estimate performance at the item level. The closer the forecasted values are to the actuals, the more accurate the model's predictions are and the better the estimated performance.