This is related to point 7, solution in search of a problem. I myself have been guilty of this when I wanted to use deep learning models just because I could. M

This is related to point 7, solution in search of a problem. I myself have been guilty of this when I wanted to use deep learning models just because I could. My much more experienced boss gently dissuaded me and I ended up with a 'boring old' logistic regression, which was completely adequate for the job. Another time I was working with an engineer who built a neural net to predict something. Turned out it was a really poor choice as interpretability was important for the problem and the neural net's predictive power was actually worse than more traditional models.