Skip to main content

Ski IQ Metric Average Calc?

Feature RequestsVideo coaching

Looking at the metrics breakdown for each segment run, it appears that the overall value of the 8 turns is not an average or a median score. It appears that your overall run % value for each metric is solely based on your 5th worst value out of the 8 turns? How is that an average? Or median? Each metric description states based on average? 5th worst score is not an average and is not a good statistical method. Let's say you score 100, 100, 100, 90, 80, 78, 76, 75. Should 80 be your average for that metric? That is what is shown based on the current programming. Actual average is 87, and median is 85. Please explain? Not sure why explanations state averages when that is not even close to what is being shown? Looking forward to explanations. Thanks Dan

Posted by Dan Kowaleski ·
1 comment

Log in to comment and vote

Comments1

  • John Moore

    •

    Jan 21

    The median value for an even number of samples can be calculated using the average of the two middle samples, the higher value of the two middle samples, or the lower value of the two middle samples. Taking the average of the two middle samples is the most common approach, but the other approaches are also valid, depending on the use case.

    We use the lower value of the two middle samples in this case because it maximizes the accuracy of Ski:IQ when rigorously evaluated across the entire population of users. Because we calculate the Ski:IQ based on your top 8 turns of a segment, this has a tendency to inflate scores. Calculating the median for each metric with the lower value of the two middle samples helps to mitigate this inflating effect while still harnessing the benefit of automatically finding the top 8 turns for the user.