This week is our last week of training and it is a dashboard week. In a nutshell we will be given a data set in the morning, then we will have around 5 hours to clean the data, build the dashboard and write a blog about it (sounds super easy right?).

So today we were give the data from the New Orleans Police Department body worm camera data to visualise. I have decided to look at the body worm camera usage compare to the complaint the NOPD received and the use of force by NOPD. Initially I have downloaded the data and load them into Tableau Prep for data inspection and any data cleansing step. Initially the body cam data has a very granular level of details hence the file size is really big, so I have removed lots of unneeded columns. When I pulled in the complaints and use of force data, I found out that the granularity are quite different, therefore I did not do much data transformation for each of them other than just cleaning up the columns and filtering out some data. The Prep flow looks very simple:

As the granularity is different, joining them would cause the data to multiply exponentially which is something I did not want to do. So I have chosen to do a relationship model in Tableau desktop itself. In that way I have kept the granularity of each data source but also avoided to explode the data massively. I know some people might not like Tableau relationship as it is not the most clear and comprehensive method for data model. However I believe it was the best way to keep myself away from the rabbit hole of configuring granularity whole day. My relationship model is as below:

After that the dashboard is done based on comparing the Body Cam usage to Complaints/ Use of Force of the NOPD in this dashboard.

The dashboard itself is very simple, I guess I could not have make much out of it as we only had 5 hours or less. And not to mention we spent maybe 20 mins tried to download the data (yes the file is huge). So overall we can see that the complaint cases has a negative correlation between the body cam usage. To take it forward I think we could have look it stop and search, allegation data as well to make this dashboard more complete.