Snapchat Data Scientist Interview Experience — Unexpected Regression Questions and Camera-Event SQL

The recruiter told me that the technical screen would cover only SQL and statistics and probability.

I could not find any new interview reports from this year on the forum. Reports from previous years covered dice probabilities, SQL-finding questions, and similar topics. The questions I actually received were completely different from what I had prepared, so I failed... I am sharing them to help people who interview in the future.

1. What are the assumptions of linear regression? How do you test those assumptions? What do you do if an assumption does not hold?

2. If every data point in a linear regression sample is duplicated once, how does the coefficient change, and how does its standard error change?

SQL

1. Given events for users opening the camera, find, for each operating-system type and device, how many users had a camera-opening event when they first launched the app each day.

2. Find the average number of times the camera was flipped within each session. The table records whether the camera was facing front or back, so this was a matter of using LAG to count how many times it switched.