Today, we were tasked with creating a fictional company and using ChatGPT to build a dataset of at least 25,000 rows so that we could produce a KPI dashboard in Power BI.
Looking back at my efforts this morning, I definitely spent too much time working with ChatGPT to create a dataset that met all the requirements and had interesting trends. For anyone who is looking to build a dataset with ChatGPT, I would definitely recommend asking it to first create a sample of 5-10 rows so that you can see the output before asking it to generate a Python script which you would then have to run in a console.
As someone who is much more familiar with Tableau, I discovered a few frustrating limitations with using Power BI. When trying to visualise customer acquisition over time, I needed to create a calculation which displays each customer's first order date. In Tableau, this could simply be done using a FIXED LOD (click here to see a blog I previously wrote on this), however, this isn't quite as simple to achieve in Power BI. My initial instinct was to create a quick measure in order to calculate the first order date per customer, but then later found that you are unable to drag measures onto an axis in Power BI. The correct way to achieve this would be to create a new column for the first order date in Power Query.
Overall, this was a challenge I enjoyed and will certainly use ChatGPT to create datasets in the future.