I build such notebooks and an important part of that is figuring out what's interesting about the data in the first place, because it's not always obvious what you want to measure.
If you look into social media analysis, for example, there are a great many different approaches being developed and documented in academic papers and codebases, but little consensus on exactly what to measure and how. Lots of work was done on follower/friend relationships only for those to be subsequently discounted in favor of behavioral patterns. Of course, the fact that the mechanics of social media platforms are opaque boxes and are competitively manipulated for a variety of purposes complicates things further.
If you look into social media analysis, for example, there are a great many different approaches being developed and documented in academic papers and codebases, but little consensus on exactly what to measure and how. Lots of work was done on follower/friend relationships only for those to be subsequently discounted in favor of behavioral patterns. Of course, the fact that the mechanics of social media platforms are opaque boxes and are competitively manipulated for a variety of purposes complicates things further.