Woah, long afternoon. I’ve struggled with CartoDB as never used it so hard but I have something.
Firstly, what I have at CartoDB are two tables:
- The table stations with the stations that appeared on 2008 records with the combined id for USAF and WBAN.
- The table observations with data aggregated by month with all the measures.
So the task is to join both tables and produce different year aggregated maps. But tey have to be representative, any station with less than 300 samples per variable has to be discarded. That measure is not perfect because I haven’t added a counter for every variable but it will suffice.
I’ve placed both tables as public so anyone can access and play with them. Well the observations dataset is not going to be too much there as I will retire it to go down to the free account type.
Once the data is loaded properly, next phase is creating the visualizations. This phase is still a mix of data and design as the data to be rendered still needs to be defined. At this moment I’ve set up four layers for the visualization.
- Maximum temperature registered
- Minimum temperature registered
- Maximum temperature range, that is the difference between the maximum and the minimum temperature over the whole year.
- Number of tornados registered.
The data are more or less all very similar to this query:
On this query first the observations data is aggregated to obtain the number of samples and the maximum temperature, afterwards a
select prepares the data to be presented, joining the data with the stations geometries, filtering those without enough samples and finallly ordering the data so the higher stations are rendered later.
Then some visualization decisions are made, selecting to start one of the palettes offered (all well known Cindy Brewer work). Then the other minimal parameters are set up on the wizard to finally fine tune the work on the CartoCSS editor.
Finally the infowindow both for hover and click events is designed, using defaults at this moment.