After a discussion with a friend about how accurate was, compared to, i set out to test it. DATA > DOGMA!

I wrote a small perl-script that scraped the temperature predictions from and every hour and saved all the values. To have something to compare to I recorded the temperature outside my appartment window as well as scraped another temperature measured about 10km away from from

Since updates their forecasts several times a day i wanted also wanted to see if the forecasts became more and more accurate. So I created a graph that plotted the differences to the measured temperature and compared it to three forecasts, one 1 hour old, one 1 day old and one 2 days old.

Since i couldn’t be bothered to compare the data I got from to the forecasts from the results was inconclusive 😁But the accuracy from was close enough for me to switch to it.