Nitpicking about an otherwise well written post: why show a kernel density estimate instead of a simple histogram? It can be normalized if you don’t want reveal

Nitpicking about an otherwise well written post: why show a kernel density estimate instead of a simple histogram? It can be normalized if you don’t want reveal the exact number of samples. Calculating a kernel density estimate just seems to add an unneccesary step between raw data and the visualization. It can hide some shortcomings in the data (e.g. low number of samples), but doesn’t add anything.