Categorizing an optimization problem: min_{x, y} 1^T y s.t. x - My <= 0; c^T x > 100; x<24; x is real 10-dimensional; y is binary 10-dimensional Select all that is true: a> Convex b> MIP c> quadratic d> Can be solved using simplex
You will be improving the function you developed prior to interviewing and will be thrown a lot of curveballs at the whim of the interviewer — because you won’t actually run your function or work with data be expected to be asked vague and poorly worded questions. This person will likely be extremely junior and have little tenure at the company or as a data scientist but definitely smarter than you — they may have taken a coursera course on modeling — so don’t expect to be able to prepare or get an offer as its extremely arbitrary and has nothing to do with your experience or ability to add value other than writing perfect code on the fly while being interrupted with vague questions with apparently very specific answers.
I had a two-stage interview, one technical coding level with some SQL and Python conceptual/coding questions. The second step was interviewing two people. Problems with the interview:
Let p be a primal number. Show that there are no more than 2 numbers m,n that satisfy m^2 + n^2 = p
how many joins in sQL
They essentially sent a non labeled regression problem
What is a confidence level?
Straight forward question of predicting a variable given the data set. Catch here is that the data set is dirty, the dataset is nuanced and there's very little time. I got the feeling that the objective here was to overwhelm the candidate to see how they'd do.
how you define accuracy ?
Interviewers doesnot know anything other than classical ML .
Viewing 211 - 220 interview questions