- trying to parse and visualize data in the same effective way as Fakta o klimatu
- while at it I noticed some weirdness when comparing the trend in Brno & Prague, thus imported all places and tried checking a bit.
Charts per type
To be able to compare the places / cities, each measure for each.
The measures are the same as on Fakta o Klimatu
- tropical days - number of days in a year where the temp reached above 30
- ice days - number of days where the max temp was 0 or less lines represent:
- "klouz" - moving average 5 years back
- "trend" - an attempt to fit the yearly values with a simple linear regression
- "lch" - since the fitting doesnt feel great for all places (looking at you Brno:), I wanted to try and check the trend since Brno got a different machine..
Clicking on the legend allows hiding places (doubleclicking hides everything but).
Charts per place
Trying to reproduce the great chart from Fakta o Klimatu. Description applies as per above, shows also the "source" data - the yearlyt values (there have been rather wild outlier years)
I tried to follow the approach suggested by Bennet Meyers, where the fitting is tried for randomly ommited values. Given the very odd years the differences are quite large, but some sets are outstanding (Brno).
- rmse - standard deviation of differences (residuals)
- mbe - mean of residuals