If we look at the source data we loaded at the beginning of the recipe, it is clear that there are some outliers present. To exclude these and get a real picture of the normal, average amount of calls, we remove the top and bottom 10 percent of the value. This is done using the fractile function. The fractile function calculates the cut-off point for 10 and 90 percent, based on our data. The Aggr function is needed because fractile is an aggregation function nested inside another aggregation.