Today we were tasked with creating a dashboard analysing the water temperature in the gulf of mexico. The data was generally nicely formatted; the only bit of prep required was combining the date field with the time field. (I didn't end up using the time element, this is probably where this planning thing Andy keeps mentioning comes in!)
From the brief I thought the temperature measurements were related to a weather event, so I wasted around 30 mins going down the wrong path before rereading the data source. I then decided it would be a good point to start planning out the charts I wanted and what questions they were answering. My first idea was a drill down heat map from decade to year, as this allowed a smaller chart to give a better overview of the data. This caused some problems the other times i've done a similar chart it was with much nicer data but got there in the end. I think it works well to show the change in the monthly temperatures and the spread of warmer temperatures as times gone on.
I was quite conscious that the sea and other bodies of water would change temperature considerably throughout the year this combined with variation in when measurements were taken meant that the view could easily be skewed. My solution was offering several filters on location and the organization who took the measurements as they seemed to operate in relatively defined zones. Combined with a month parameter, so only the same months were being compared.
I presumed that there would likely be a relationship between global emissions and sea temperatures so i found a data set that detailed co2 emissions per capita and then blended that with the water temperature dataset. I then created a dual axis bar and line chart to see if there was a correlation which, there seemed to be.
One measure I thought would be interesting was the range of values across a given month across the decades, i thought a box plot would best detail this well, as well as showing the change in the medium temperature it also shows the range of temperatures and whether there are more extremities as time has passed.
The final chart was a map showing where the temperature was recorded, this was going to be filtered by the other charts, but i ran out of time to get that configured.