Analyzing Data When p >> n

MediumClassical ML

Reported in interviews at Citadel

Problem statement

How would you analyze a dataset with very few observations and far more features? Explain why standard methods fail, which alternatives you would consider, and how you would judge reliability.

Show hint

What does least squares need from the design matrix in order to return a single well-defined answer?

Relevant roles: ML Engineer, Applied Scientist, Data Scientist and Research Scientist.

Related free problems