Some technical questions from my Facebook DS interview:
How do you use logistics to test coefficients in ML?
How do you deal with skewed data in ML?
Compare logistic regression and random forest in ML.
What is regularization in ML?
You are given a user table with the data values: userid, amount, distance_from_home, and fraudulent. What models can you use to determine whether there is fraud? How do you determine whether your model is effective?
Moving onto the final round soon. Fingers crossed!