Lyft/Uber Metric Interview Question and Answer: Tips for Data Science Interview Success!

This video provides a sample answer to a real data science metric interview question from Lyft/Uber. "We have a dashboard tracking our metrics, and average ETA is up by 3 minutes. How would you investigate this problem?"
Cracking Metric/Business Case/Product Sense Problems
• Crack Metric/Business ...
🟢Get all my free data science interview resources
www.emmading.com/resources
🟡 Product Case Interview Cheatsheet www.emmading.com/product-case...
🟠 Statistics Interview Cheatsheet www.emmading.com/statistics-i...
🟣 Behavioral Interview Cheatsheet www.emmading.com/behavioral-i...
🔵 Data Science Resume Checklist www.emmading.com/data-science...
✅ We work with Experienced Data Scientists to help them land their next dream jobs. Apply now: www.emmading.com/coaching
// Comment
Got any questions? Something to add?
Write a comment below to chat.
// Let's connect on LinkedIn:
/ emmading001
====================
Contents of this video:
====================
0:00 Intro
0:42 Lyft Metric Interview Question
3:36 A Mock Interview
10:18 Interview Summary
11:29 Outro

[@clapdrix72]

You could save yourself time by looking at the ground truth vs ETA first. If the ground truth was affected then you can safely rule out other possible causes that involve some data or algorithmic related to ETA generation thereby increasing the likelihoodyou solve the problem quickly. If I were interviewing I'd probably take off minor points for not identifying the most efficient order of exploration. Aside from that it's a very thorough and well thought out response!

[@jeoffleonora4612]

This is the exacting video I have been waiting for about real metric interview question. Thank you for making the video Emma! It's really worth waiting for a couple of times. And I'm taking notes!

[@sparklingamber792]

Love all your videos! Improved my performance during the product sense interview a lot!

[@brothermalcolm]

Awesome job, this complements the exponent's mock interview video really well!

[@tjc7524]

Great job! Learning tons of interview knowledge from your series. Thanks Emma!

[@heisenberggiao9303]

This is a great video, it helped my Uber interview a lot! Thank you Emma!

[@MatthiasEkundayo]

Thanks a lot for making this video! Really helpful!

[@akshaychauhan2757]

Thank you so much for this video. Please keep posting such informative stuff.

[@amneymnr6455]

Thanks Emma! I am interviewing for FB next week and this is so helpful!

[@wisdomhaijiang2004]

So good! Doing very well, Emma!

[@cooltrucly]

I'm a student, and I have my very first data science interview tomorrow! Watching this video was super helpful since I had no idea what to expect before. Thank you, Emma, for presenting such an organized and clear video to help people prepare-- most importantly, for free. Just subscribed, looking forward to seeing more :).

[@emma_ding]

Glad it helps. Good luck with your interviews!

[@amirkhan355]

You're awesome, I love you!

[@zdaman011]

Another super helpful video. Thanks, Emma!

[@emma_ding]

Glad it was helpful!

[@Nedwin]

Thanks for this Emma! ❤️

[@mohdtanveer3092]

Thanks for sharing this. Looking for more of these.

[@yidanhu7889]

Also, thank you for your effort Emma. This video is really help!

[@emma_ding]

Glad it was helpful!

[@mayavaramthamizhachi]

Thank you so much for sharing. It is super helpful!

[@niveditakumari701]

Amazing Video, best content I found on internet on product Sense Interview

[@nates3361]

Thank you so much Emma, this was super helpful

[@alexz7032]

This is awesome. Thanks!

[@SoumyaDasgupta]

Great effort helping noobs like us. amazing.Thankyou so much.

[@281019641]

Thank you for the video. This video is so helpful as it gives a real sense and direction of how to answer interview questions.

[@emma_ding]

Glad it was helpful!

[@songxiyou2347]

It def helped!! That's the video I will watch for 10+ times. It's what I am looking for during interview preparation but i didn't see any other you-tubers are doing!! Thanks Emma, I will follow up with some product questions that I feel hard to answer later. Looking forward to more videos like this!

[@emma_ding]

Glad it was helpful!

[@songxiyou2347]

@@emma_ding Hi Emma, this is an interesting question I met during the interview preparation, would be great if you could give some instructions on the answer structure, I can't tell if it's a metric selection problem or n A/B testing one. { We are launching a new driver app with a better UI. The goal is increasing driver earnings by increasing their number of trips. Outline a testing strategy to see if the new app is better than the old one } Thank you and happy holiday!!

[@takozhangsbox3414]

Thank you very much for your effort. It helps a lot.

[@emma_ding]

Glad to hear that!

[@mahdimerced]

As always, it was useful

[@emma_ding]

Glad to hear that!

[@sofiagreen4348]

Great info! Do you have any advice videos for sale/revenue forecasting metrics interview question?

[@slapfighting]

I have faced multiple questions like this & it gave me a good idea on why i was failing such interviews

[@emma_ding]

Glad it helps! More to come!

[@yuanman2012]

Thanks for the perfect answer. Could you tell what is the expected interviewee level could give this perfect answer?

[@jiehu1337]

Thanks for sharing Emma, at 8:20, if you would continue to dive in, how would you proceed since you have already explained it.

[@tongkezhao2188]

Hi Emma, thank you for the video! High quality as always. Quick question, will it be good to first dive into various segments after looking into outliers? I found a article mentioned to dive into ‘high-level’ reasons first, such as regions, various time periods, etc. Got confused here. Thank you!

[@emma_ding]

Sure, the order does not matter too much as long as your answer is comprehensive and practical.

[@hezhaojiang3525]

Excellent video. Would love to see how you crack a A/B testing case. It must be interesting and helpful if you could focus on answering very typical A/B testing tech questions. (How do we define success. How do we design an experiment. What’s the sample size. )

[@emma_ding]

Here you go kzbin.info/www/bejne/jmnYZ56oacurjsU!

[@yifanxu9777]

this is a good video for diagnosing problems Thx a lot

[@emma_ding]

So nice of you! Glad it was useful!

[@katekatebangbang2435]

We can also use extreme value as third quantile + 1.5IQR, any value > than the threshold can be consider as extreme value :)

[@emma_ding]

Yep, another good way to filter out outliers!

[@yalili1600]

Could we get a copy of the PPT of this if possible:? It would be awesome.

[@SonuKumar-gt5xs]

Hi Emma, this is really helpful. Do you have 3-5 similar questions for practice, thanks

[@emma_ding]

Hey Sonu, thanks for your question! 😊 I'd love to offer you a link www.datainterviewpro.com/#free-resources which grants you free access to my Product Case Interview Cheat Sheet. This resource contains some sample questions for you to practice. Hope this helps!

[@yidanhu7889]

Hi Emma, I recently had an interview with a problem solving question. I tried my best to structure my answer and find the reason for all the aspects. I did not ask the interviewer if my answers made sense because the interviewer looked nice and for the points I answered, she nod multiple times and said "I like this approach", "good point" which made me think my answers were good. After a day, I got an immediate rejection which surprised me. The kind recruiter told me the reason is I only investigated the problem but no solution/next step/action item was mentioned. I do not fully understand the feedback. 1. The question I got was "please investigate the problem xxxxxx". If I also need to provide solution after investigating, why the interviewer did not tell me? she can just ask a follow-up question, right? 2. The appraisal I received during the interview did confuse me. If the interviewer was not satisfied, I believe she should ask a follow up question to guide the interviewee instead of appraising me. 3. Do you think the interviewer's behavior is normal? If so, can I summary that "Do not believe all the feedback you heard during the interview, please still go ahead and confirm with the interviewee if I need to dive deep."? Hope you can share your comments/thoughts. Thank you in advance Emma!

[@emma_ding]

Hmm.. it's a bit difficult for me to provide feedback without knowing the context. But in general, every interviewer has his or her style. Some interviewer might have pretty high standard and wouldn't want to "give away" the answers through feedbacks but still want you to have a pleasant experience. My only suggestion would be try to explicitly confirm whether your answer is align with the interviewer's thoughts in order to get on track. Good luck to your future interviews!

[@yidanhu7889]

@@emma_ding Thank you for the suggestions Emma!

[@byronhan8961]

I had the exact same problem. The biggest takeaway I got for my next interview is "always always always" checking with the interviewer constantly, there is no way to just go ahead talking when the interviewer is nodding and smiling.

[@yidanhu7889]

@@byronhan8961 strong agree!!

[@linhe5896]

My first question when I saw this video title: did it breach NDA? I am glad Emma you didn't go down that path and proactively mentioned 'make sure you don't break NDA clauses'. Phew! Another well done video, I like how you laid out the framework and twisted it a bit during your mock answer.

[@emma_ding]

This question is already publicly posted on glassdoor for a long time. :)

[@nehamishra7442]

Hey could you please do one video for FB's product sense question for Data Science?

[@emma_ding]

Here you go kzbin.info/www/bejne/gHSxpHxmidmBfdk!

[@yueedwards1]

Hi Emma, how long is it generally acceptable for the interviewee to take a pause to gather thoughts before answering questions during an on-site interview?

[@clapdrix72]

As a Sr. DS at a major tech, I regularly interview DS candidates in the tech portion. Take that as you will for the relevance of my opinion. It always depends on the interviewer but for me it depends on the difficulty of the problem. For intentionally challenging "case study" style questions, rule of thumb is not more than 2-3 minutes of silent thought. If you need more time just ask a clarifying question (but don't make it obvious you're stalling), or better yet walk the interviewer through your thought process and identify what the specific challenge is your trying to work out. They care more about HOW you think than WHAT you know so if you're a bright, systematic thinker you'll get brownie points even if you don't get the right answer. On the other hand, if it's an easier problem or one with a simple, objective answer then a long pause is viewed negatively cuz you basically either know it or you don't and if you can't jog your memory in 10-20 seconds you just admit defeat on that one and move on (but don't lose confidence - or at least don't show it).

[@cindyzixinzhou3438]

Hello Emma, thank you for sharing, really helpful for me! I have a question that when consider the time aspect, is it better to ask interviewer: is this a sudden or prograssive change? first, so that we can go the correct way directly. Or as your video to analyze both cases, then ask if there is any situation happening? Or we can ask in this way: is this a sudden or prograssive change, or you'd like to know how I think in both cases? (my concern is your analysis is more comprehensive, which can show the thoughts. but it may spend more time.) Also, I am confused, after give the analysis in both sudden and prograssive change senerio, you ask 'if there is any situation happening?' In this way, is our goal in this listing process is trying to list the possible reasons in a structured way, but not confirm if it's a sudden or progressive change? One more thing is that, when ETA is different from the ground truth, your suggestion is we can further investigate if anything such as a recent product launch changed the calculation of ETA. I cannot get how to connect calculation of ETA to the product launch. Can you explain it further? 上述我不确定英文是否表达清晰，因此用中文重复一遍。第一个问题是：在面试中，是直接先问is this a sudden or prograssive change?去norrow down the question, 还是按照你视频中的，分别分析两种情况下可能的原因，再问是否因为其中某种原因；亦或是换一种问法：is this a sudden or prograssive change, or you'd like to know how I think in both cases? 第二个问题：因为在视频中，你是分sudden和progressive情况列举了可能的原因，然后问if there is any situation happening?，以此，这一过程的goal是为了让列举更structured，然后看那些列举原因中有没有是cause，还是确定是sudden还是progressive。（因为metric change只有可能是sudden或progressive，因此如若像视频中面试官的反馈-列举原因都不是cause，我们可以排除掉列举原因不是cause，但无法确定metric的变化是sudden还是progressive；那么在做segment时，是否有前提假设：这个metric是sudden还是progressive，还是这个假设不重要，我们的segment analysis不需要time这个假设） There are three questions, I hope it makes sense，thanks！

[@cindyzixinzhou3438]

A plus for the third question, why you are gonna to check the computation of ETA when ETA and ground truth are different?

[@emma_ding]

​ @Cindy Zixin Zhou I recommend asking "was it a sudden change or a gradual change" before answering the question.

[@tejasphirke3436]

Nice video 👍 Can you make a video on what projects should be in resume to get the job?I m confuse about to make projects? As a fresher it is a kind a hard to find job.

[@emma_ding]

That's a good idea. Stay tuned!

[@rubyjiang8836]

Can you describe what’s the difference between ETA and ground truth. I’m very confused here... thanks!!

[@emma_ding]

ETA is estimated time of arrival while ground truth is the actual time of arrival.

[@rubyjiang8836]

@@emma_ding oh now I understand Thank you!

[@XingchangCheng]

666666