1: Trees-based methods are sometimes called greedy algorithms. Can you explain what greedy means, and then explain what is a potential downside of using a greedy algorithm? 2) Can you explain the concept of multicollinearity and how it can impact a model? Then discuss how you measure the amount of multicollinearity. Finally, using the measure you identified, how much multicollinearity would cause concern for a model, and why? 3) A colleague approaches you with an idea to run a marketing mailer experiment. The colleague wants to test the impact of two different marketing mailers (let’s call them Mailer A and Mailer B) on life insurance sign-ups. The colleague asks you: “how many mailers do I need to send out to have a statistically significant sample?” How do you respond? What questions would you ask the colleague to help answer their question? 4) Assume you deployed a binary classification model and after 12 months your business partner asks you to verify the model is still performing as expected. What steps would you take to validate the deployed model’s performance? What different recommendations would you give your business partner if the model is performing as expected versus if the model is showing deteriorated performance? 5) Please share what you like about the open role. 6) Can you tell us about your prior mentoring or leadership experiences? 7) why statefarm .
