The purpose of this is to calculate the moving average by the Holt-Winters method, a triple exponential smoothing model where the moving average depends on the window size and the alpha, beta, gamma∈(0,1]. alpha, beta, and gamma parameters default to 0.3, 0.1, and 0.3, respectively. Holt-Winters can track seasonal aspects of data by using addition and multiplication. The corresponding setting types are add and mult. The rest of the aggregation for the Holt-Winters model is as follows:

"weekly_moving_avg": {"moving_avg":{"buckets_path":"weekly_change","model":"holt_winters","settings":{"type" : "mult","alpha":0.8, "beta":0.3, "gamma":0.3, "period":1},"window":4,"predict":4}}}}

The following chart includes all the supported moving average models corresponding to the aggregation examples in the Moving average aggregation section of this chapter: