On Thursday, the fourth day of Dashboard Week, we received the Crossword dataset. So far, we had alternated between Tableau and Power BI, and on this day, we were required to build the dashboard using Power BI. At this point, I was already pretty comfortable with the tool—but still, you never know!
Getting Started
I haven’t played crossword puzzles in a long, long time, so this dataset was quite interesting to me at first. The data itself had already been prepped and transformed by the Coach beforehand, so I didn’t need to do any heavy cleaning. Most of the minor adjustments were made in Power Query.
This time, I didn’t struggle to understand the data, but I was puzzled about how to approach it. I spent a good amount of time experimenting, trying to see what kind of narrative I could pull out of it. Working with a dataset that contains a lot of text really made me want to learn Python for text analysis. Although Power BI does support some of that, I decided to save it for future projects. Besides, it’s not like the dataset only contained text.
I went ahead and created a bunch of summary charts in Power BI to establish a solid foundation for any potential story direction.
Start Building
Eventually, I decided to focus on clues. I started off with one big question: How many clues have been used in crosswords? I know—not the most exciting path—but it was enough to spark an idea. I noticed that “Girl’s name” was a favorite clue among editors/authors for quite a while, so I decided to go down the rabbit hole and focus solely on that theme.
Final Thought
In the end, I decided to double down on the “Girl’s name” angle mainly due to time limitations—we only had seven hours to complete the entire dashboard. While it wasn’t the most exciting or groundbreaking direction, it gave me a focused path to work with and allowed me to make solid progress without getting stuck exploring too many ideas. This day was more about making quick decisions and sticking with them, rather than chasing the perfect insight.