Looking at the raw data sources, I realized, that for my location (Hamburg, Germany), the data on humidity from gfs really seems to be an outlier sometimes, aff

Looking at the raw data sources, I realized, that for my location (Hamburg, Germany), the data on humidity from gfs really seems to be an outlier sometimes, affecting the forecast.io-mean. Maybe, just maybe the some other statistical measure would compensate for this and make the forecast a little bit more accurate. Median, or something like trimmed mean. And, as the service is using the location of the user, why not show the weather in °Celsius, if the region, where the user resides measures in °Celsius (and Fahrenheit, where Fahrenheit is used). I really had to look hard to find the button, to switch. But non the less, great service, design, that I like, not too much info at first glance, but the most important one, imho.