Daily weighting is made up without any real insight into operations. Furthermore, whatsthefare says that it bases its own approximations on relatively sparse data from user's use of their systems, not realtime monitoring on a service that's probably prone to constant change.
I think I mostly find it incorrect to essentially take a pin-hole view at the data through 5-6 averages, and extrapolate it towards a somewhat biased result.
Even if you drastically change the daily weightings you don't move the overall output very much at all.
As for the other data they describe their collection methodologies, and the numbers they produce are in line with trends that Uber has indicated. You're certainly free to produce a dataset of your own but until you do that this one doesn't seem especially bad or anything.
I think I mostly find it incorrect to essentially take a pin-hole view at the data through 5-6 averages, and extrapolate it towards a somewhat biased result.