Couldn't agree more.
Yesterday I started an analysis with the goal of just producing one plot, as my manager asked.
After that, I thought: cool let's create a facet plots for better clarity.
Ohh, let's plot a boxplot, can see the percentiles.
Oh that's nice. Why is the median of variable A more than B? Let's find out.
And like this, the whole day went by.
Sent the final report and findings to boss. And he liked it.
Without random exploration, Data analysis is often incomplete.
