5.2.2 Timeseries Algorithms Supported

This topic describes the information about the timeseries algorithms supported in the framework.

By default, the framework uses Exponential Smoothing to forecast from timeseries data. It evaluates 14 different algorithmic combinations to best fit the below patterns:

• Error type (additive or multiplicative)
• Trend (additive, multiplicative, or none), including damped trends
• Seasonality (additive, multiplicative, or none)

Note:

The user is not required to select any algorithmic combinations. The framework evaluates and selects the best fit combination.