Today’s dashboard is using American Candy Survey data (2014 - 2017) conducted by Science Creative Quarterly (SCQ), which immediately indicated to me that there was plenty of restructuring to do. We’ve spent two days of training with Sam working with survey data before, so we were well prepared.
We weren’t prepared for the twist: 1) we can only use Tableau Prep in the browser to do data prep and 2) the dashboard must be built in web edit.
Unfortunately DSNY2 experienced multiple periods of the server stalling, so I’ve begun writing my blog post early (almost lunch time here!).
We’ll see if I can get a dashboard together by the end of this blog post and by the end of the day… (Spoiler, I was able to.)
There were a few caveats with the data - the raw data for 2014 is missing and has a different structure; 2015 data does not have the “MEH” option (the scale goes from despair, meh, joy).
I started a prep flow using data from all 4 years that took until lunch, but didn’t run due to multiple reasons - there was an issue with the 2017 data (that DS35 alluded to in their blog posts as well).
After lunch, I started a new prep flow that used 2014 data as the base - it was aggregated compared to the other years, and so I restructured data to match it. The reason for that was that SCQ uses a metric called “net feelies”, which is the number of Joy responses minus the number of Despair responses, which was present in the 2014 data.
I started with a short description of the data and the metric “Net Feelies”. Obviously, a lollipop chart was the most appropriate here (strawberry pink, for Coach Ann’s favorite flavor) that shows the Net Feely across all years and all items (yes, ITEMS) in the American Candy survey.
I grouped items by Candy, Not Candy (but still edible), and Not Edible, and looked at the average Net Feelies as the years went on.
While cleaning the data, we verbalized strange things we saw, and I wanted to explore that more with the Not Candy and Not Edible items with their Net Feely score:
ONE MORE DAY.
