Sample k items uniformly at random in a streaming fashion from a list of unknown length.
Data Scientist Interview Questions
33,874 data scientist interview questions shared by candidates
Case study
10 questions to do in 2 hours.
3 questions, two sql+ 1 python (though can be also easily solved with sql). The sql question is about finding the most popular track and for each track, the day it had most popularity. the python question is about adding certain advs. seconds on every 3,5,15 stream a user had. The interviewer appears having limited experience with sql, as both interviewers had moments first doubting my code, then admit they made mistake. As one example, when doing an inner joining of two tables I had 'From tableA a, tableB b where a.metric=b.metric, the interviewer first asked me where's my inner joining of the two tables, as in her mind it should be written as 'From TableA a inner join TableB b on a.metric=b.metric'. After I explained, she realized both ways writing are equivalent. Though I got all three questions right and well explained, I did not pass the tech interview and there was no feedback.
Given a cube and a random walker on the cube, find the expected value of the # of steps that the random walker takes to reach the opposite vertex of the cube.
If a jar has X red balls and Y blue balls, what is the minimum number of draws that is necessary to ensure that you have one ball of each color?
Technical Interview Questions: What are the assumptions for Linear Regression? What is a Random Forest? How is a decision tree built?
What’s the probability of a type 1 error. What happens when the sampling distribution is altered (e.g. filtering out all values below the mean) and how does it effect type 1 errors.
Write a program in Python to calculate tax if Salary and Tax Brackets are given as list in the form [ [10000, .3],[20000, .2], [30000, .1], [None, .1]]. You don’t know in the beginning how many tax brackets are there. You have to test for all of them.
Given an event-level table of interactions between pairs of users (note that there aren't duplicates in one day for one pair of users), for each possible number of "people interacted with" find the count for that group in a given day (i.e. 10 people interacted with only one person, 20 with 2, etc.).
Viewing 131 - 140 interview questions
See Interview Questions for Similar Jobs
Data EngineerQuantitative ResearcherData ModelerData AnalystStatistical Modeling AnalystAnalytics ManagerData Mining EngineerMachine Learning ScientistAnalytics ConsultantBusiness Intelligence AnalystResearch ScientistData ScienceQuantitative AnalystMachine Learning EngineerData ConsultantModeling EngineerData Mining AnalystStatistical ConsultantComputational Biologist