I don't see how this can be proven, given that weather forecasting itself is highly error-prone, so how do you discern poor quality due to government forecast from poor quality due to the inherent nature of the field itself? I've seen nothing from Accuweather that makes me think they are a marked improvement over the NWS.
I can think of one marked improvement: one costs the taxpayers nothing, and the other does.