How deep should one be acquainted with theory in order to be a decent analysis?

theWay

In my data mining class, we are learning the most common supervised learning techniques so far.

For example, if you randomly select a successful analysis in industry, would he/she be able to derive the maximum likelihood iterative process of Logistic Regression without having to review any textbooks?