The calculation of the median includes zero values, which is correct mathematically, but it creates the impression that "typical" performance is lower than the mean. Should we offer both the median (with zeros) and the median (of non-zeros) to give users the full picture?
That number looks wrong.
Why is the week-start date off by one day in some years? Is there a leap-year bug?
Confused by the 2020 value — is that an outlier?
Balin Boffin The axis range is too compressed — could we zoom in?
Decimals are inconsistent between Iberia and South Africa.