Healthier Candy Choices for the Maven Halloween Challenge

I already had an angle for where I wanted to take this project prior to looking at the data. The goal was to prioritize choosing the the healthier options coupled with them being the most popular according to respondent feedback - really simple, to the point where there were times I felt I was doing a disservice to the project and the data. I always find something an any EDA I do that has little to do with the outlined question but is still significant in some way. In this particular project I found that there were a number of name/ownership changes and some treats were even discontinued. This was possible because a lot of the work for this project happened outside the table due to the need for additional information.

At the end I managed to answer my questions, they were informed/guided by the data from the table, and I think my motivations were sound.

Below is my notebook for the project done in Databricks Community Edition where I also included more of my rationale and provided some links to resources used throughout.

View Fullscreen

I definitely could have taken this project much further along the path I chose. One way would be constructing an additional table that had all the nutritional data I found so I could also quantify some of my justifications for decisions I made, but I feel it was enough (with the criteria I had at the start) simply to rely on the presence/absence of certain elements.

A big lesson I took away from this project as the need to remember not every challenge needs the most complicated/impressive strategy or solution. AI or algorithms, linear regressions and the such. Sometimes it’s a simple yes or no and expanding a little on that. I had questions I managed to answer and that’s enough.