The second day of dashboard week began with each of us presenting our visualization from the previous day (here) and explaining the process of how we arrived at the final dashboard.
Our present this morning from Andy involved a data source by NOPD (New Orleans Police Department) relating to body worn camera video recordings between February 2014 and October 2018. This data set *should* have been much simpler and faster to use, being only 2.7 million rows when compared to that of yesterday which contained in excess of 252 million. However, as Andy always promises he threw in a curve ball of adding in a .shp file. With the data being in such a clean manner we were extremely thankful after all the issues we encountered yesterday. The only troubleshooting problem to overcome came in the form of joining our two data sets together, eventually working out the reason they wouldn’t was due to one side of the field we were trying to join on being identified as a string and the other as an integer. To get around this issue we used a calculated field inside the join function:
After this came the process of finding the story in the data, probably my favourite part of all! Every data set has it’s own unique story to tell, some are easy to find and some simply aren’t. The best way to go about this is just try, fail and try again until you succeed. Luckily for me I found fairly quickly a large spike in the number of video recordings taking place in district 4 for the month of October 2018, ranging between 1,200 and 2,200 while asking myself, why would this be? After a little research on the web i found a story in the New Orleans News of a man with a machete outside the city hall. I believe this could well explain why we see a vast increase in the number of police bodycam video recordings throughout October (and onwards when the new set of data is published) as there is a heightened level of risk at this moment, but the number should start to decline back to more normal levels over the following months in the 100-200 range.
Dashboard Actions: Declutter your dashboard
Today I heavily used highlight actions in order to bring together my dashboard, utilize my filters and identify the key pieces of information a user is looking for. Without applying these actions my line graph would have been close to being uninterpretable as you can see below. The left shows it prior to the dashboard highlight action and the right post.
I look forward to writing another one of these tomorrow to show you the mad concoction of data sources Andy will throw at us!