How to interview: Probe, probe, probe
you gotta dig deep
This role play illustrates how to dig deep as an interviewer.
Interviewer: Hi, can you tell us about a data science project you've worked on in the past?
Interviewee: Sure! I recently worked on a project to analyze customer behavior data for an e-commerce company.
Interviewer: That sounds interesting. Can you walk me through your methodology for analyzing the data?
Interviewee: Of course. We started by cleaning and preprocessing the data, then we used a variety of statistical techniques and machine learning algorithms to identify patterns in the data and make predictions about customer behavior.
Interviewer: Can you give me an example of a statistical technique or algorithm you used?
Interviewee: Sure, we used logistic regression to predict whether a customer would make a purchase based on their past behavior on the website.
Interviewer: Interesting. And what was the accuracy of your model?
Interviewee: Our model had an accuracy of around 85%.
Interviewer: Hmm, that's not bad. But can you tell me about the false positive rate of your model?
Interviewee: (pause) I don't have that information off the top of my head, but I can look it up.
Interviewer: That's okay, let me ask you this instead. How would you interpret the results of your model if the false positive rate was 50%?
Interviewee: Well, if the false positive rate was 50%, it would mean that half of the customers we predicted would make a purchase actually didn't. That would be a big problem for the company.
Interviewer: Yes, exactly. And that's why it's important to look at both accuracy and false positive rate when evaluating a model. It's easy to get caught up in the high accuracy score and forget about the false positive rate, but that can lead to costly mistakes for the business.
Interviewee: (chuckles) Yes, I see what you mean. I guess I got caught in that trap too.