Hello and welcome to dashboard week!
This is a week that is done toward the end of training for every Data School cohort. The week is designed as a bit of a pressure cooker situation, encouraging us to time-box and use all the skills we have learnt over the course of the training so far.
To kick things off, we are starting the week by looking at LA crime statistics.
Getting to know the data
Given the data set in the morning, my first step of the day is off course to explore the data. I first open the .csv document in Alteryx and explore the column names, and any that may be unclear to me.
Being a great team, we discuss as a group what we think some of the column headers are describing and start looking up the ones we are unsure of. I research and filter data accordingly – I had to turn to Google often to understand crime types.
Once I have an idea of the data and what possible things I may want to explore, I need to start cleaning the data.
My idea was to have a heat map where you can filter by category of crime. Using the Unique tool in Alteryx, I can see that there are 138 categories of crime. This is far too many to have in a dashboard from a UX perspective. I decided to collect the categories into logical groups so that UX will be enhanced in the final product.
Small hiccup: I thought I could use the IN() function in Alteryx instead of using Contains() with many OR operators, but unfortunately Alteryx seems to struggle checking the entire string when using the IN() function. I have therefore wasted some time altering my functions to use Contains() instead.
There was some going back and forth between Tableau and Alteryx in order to get all of the data I wanted for my dashboard.
Enriching the data
I initially took another hour to group the “Premise” data. However, I noticed with such detail in this column that there was a lot of information here. I expanded out the premises column from 1 to 8 columns. I even explored some information about the train lines, however, I ran out of time to include this in my final dashboard. Similarly, I made a filter for locations that were more likely for visitors or locals (or both). Once again though, I was unable to explore this data in my dashboard due to time constraints.
I decided to make an exploratory dashboard using the data. The target audience is tourists or people who will be travelling to LA and want to see where and what types of crimes are common for different activities.
The user can filter by gender, accommodation type, if they will be using an ATM or bank during their visit. They can also filter by activities, and what kinds of things they would like to see. If they are the sporty type, they can select several sports venues to see the crime stats and areas connected to these.
Finally, they can also select their mode of transport and see how this will affect the crime they may experience in the city.
I made a simple dashboard with these filters and at the top, the viewer can see the top 10 crimes as well as a heat map of the locations of the crimes.
Because of the level of detail that I went into in my filtering process regarding the premises column, I would have liked to have added more filters to my dashboard, but unfortunately, there was no time for this.
Of course, I would love to have a better and more detailed dashboard. But for one day, including data exploration and wrangling, I think I’ve done alright to get a product up.
See you tomorrow!