Ues res_avg to compute the median of the outliers. And the result is just the res_avg, but when the outliers aren't larger than the label, we will choose another number.
At Hexagon-ml we value math and data. We provide an ability to practice, collaborate and learn datascience.