The hiring process for Data Science positions can be quite challenging at times. With the number of Data Science jobs increasing, there arises a question of how to better assess the applicants. Hear Merishna, a data strategy consultant, shed some light on the topic.
Generally speaking, a lot of data science interviewers have the tendency to ask direct questions regarding a certain topic or an algorithm.
In theory, interviewers are correct in doing so, since an applicant should remember all of the terminologies and algorithms. However, in practicality, such questions are hard to answer given the applicant hasn’t worked on the topic for sometime. Plus there’s the added pressure of the interview itself.
Let’s discuss this with an example.
A common data preprocessing question one might ask in an interview is the difference between label-encoding and one-hot encoding. In most cases, such terms escape out of the head of the applicant during interviews when no context is given. As a result, they might fail to recollect or phrase the meaning of such terms correctly.
So, a proper way to frame this question can be by asking how the applicant might manipulate a categorical column so as to feed it into a data model. Based on his/her answer, you can further ask the reasoning behind the choice of the preprocessing technique and why they didn’t choose the alternative.
So, if their answer is suppose, to substitute the values of the categorical column by a numerical ordering, i.e., label encoding, then you can ask why they didn’t prefer to create dummy variables, i.e. one-hot encoding. This will help you to assess the conceptual understanding of the applicant rather than just testing their theoretical recollection of terminologies.
Always remember that as an interviewer our job is to assess the problem-solving experience of the applicant rather than their ability to memorise a dictionary.
Do you want to learn Python, Data Science, and Machine Learning while getting certified? Here are some best selling Datacamp courses that we recommend you enroll in: