Data Volume Definitions

Target Historical Data Points: The number of historical actuals in a target data series. A target that has monthly frequency going back five years would have 60 data points (12 * 5 = 60).

Data Set Historical Data Points: The maximum number of data points across all targets in the data set.

Training Depth: A scale of 1 to 5 of how long models will train. The larger the training depth, the more iterations of the model will run. This may lead to better model accuracy, but with longer train times.

The following graphic summarizes how the number of data points, model training depth, and number of targets, influence various components used to build and utilize models for generating predictions.