such a good video. I just missed the python code aside . Excellent video

honestly i found the response of the interviewee so messy .. are they correct ?

Good bad example case :/

Which platform are you using

I would have asked about the provenance of the data, i.e. on what grounds the sellers and transactions were classified as fraud. If these were simply reported as fraud by other users, a fair share of these could be from bad-faith competitors. In this case, I would think of alternative ways to gather data, and propose a more lenient decision boundary for fraud. The interviewee did not have time for a deep dive into the mechanisms for determining image/title misrepresentation. A subset of misrepresentation (image/title mismatch) could be captured by CV models.

why pooled variance = 2* control sample variance?

I liked the practical representation with steps in this Video. It is very nicely done but few of the very important points are mentioned by words. This may need to take up as additional one line to slides incase of Success Metrics, Launch Decision and others. Cool one.. 🙂

Very useful

Thank for this content. I am loving all the mock interviews and the story telling structure in an interview.

Where did you go?

But in case of proportions calculating variances , how we can know the proportion for the treatment group. why there we haven't taken 2 * p1(1-p1) like how we done in case of mean?

My solution for last quest: WITH cte AS ( SELECT user_id, payment_method, COUNT(*) AS frequency FROM amazon_transaction GROUP BY user_id, payment_method ), cte2 AS ( SELECT *, DENSE_RANK() OVER (PARTITION BY user_id ORDER BY frequency DESC) AS rnk FROM cte ), cte3 as( SELECT c2.*,at.* ,rank() over(partition by at.user_id order by transaction_date ) as rnk2 from cte2 c2 inner join amazon_transaction at on c2.user_id = at.user_id ) select user_id,payment_method, frequency from cte3 where rnk2 = 1

Isn't this more product management?

This was great

User clicks in the following session is not a violation of IID… your analysis unit is ‘user’… if you are only measuring user clicks uplift in the first session, how could you confidently say that once launched the feature is actually increasing user clicks? This just doesn’t make sense…

Best video for AB testing on YT. Thank you Dan 🙌

significance level is the probability of making type 1 error, which is rejecting null hypothesis when it was true

i dont understand why all you business case has to relate to ML

The definition of Power is in fact the probability of 'not' making the type II error, not the other way around mentioned by the interviewee. It is the probability of rejecting the null hypothesis when, in fact, it is false.

Another approach I'm thinking about is using INDEPENENT Z TEST ...where the null hypothesis would be that there is no difference in the proportion of heads and tails. And, if the null hypothesis is rejected we can say that the coin is not fair.

should the question be "measure the success Of youtube"?

Why can't we use MIN function to determine the first order the driver delivered: WITH zero_rating_first_order AS( SELECT driver_id, MIN(order_place_time) FROM delivery_orders WHERE delivery_rating = 0 GROUP BY driver_id) SELECT COUNT(zrfo.driver_id)*100/COUNT(DISTINCT(driver_id)) FROM delivery_orders JOIN zero_rating_first_order as zrfo ON driver_id; Or is there anything that I missed?

When thinking about the problem again, it’s almost useless. When you know the flight delay 30m before the departure, you are already at the airport😂

Guys, where can I prep for those mock interviews?

Dan we can see that we have chosen revenue_day_per_user as a success metric it will happen when the user makes a purchase so our target population should be the user who purchase right not the user who search then in this case should be the user who purchase then we will be able to get the revenue_day_per_user if we are concerned for the user who search then I think CTR can be a good metric to calculate or conversion rate that can also be good can please advise on this part because i think all the users who searches will not purchase. Some may drop at browse for item

Most useful and educative video on problem solving I have seen over the Internet! Thank you.

Dont bother watching, no feedback at the end and you will waste 16 minutes

best a/b test videos i've ever seen, Thanks!

this lady needs to improve her English speaking

Great Mock interview: for the effect of UI or algorithm we can : A/B/N Testing Design: - Control Group (A):This group does not have access to the "Call Suggest" feature. This serves as the baseline to measure the impact of introducing the feature. - Treatment Group 1 (B - Algorithm Only):This group receives the "Call Suggest" feature but with a basic or standard UI. The purpose of this group is to test the effectiveness of the algorithm itself without significant influence from UI changes. - Treatment Group 2 (C - UI Only):This group receives the same recommendations (possibly using a basic algorithm or random suggestions) but with the enhanced UI designed to highlight or promote the "Call Suggest" feature. This helps isolate the effect of the UI design on user engagement. - Treatment Group 3 (D - Full Feature):This group receives the "Call Suggest" feature with both the advanced algorithm and the enhanced UI. This group tests the combined effect of both the algorithm and UI.

You should've added that if there's no bias in *why* the 30% of x1 is missing then it's perfectly reasonable to take mean(x1)

Your face is on top of the critical solution number (z=4.0) the whole time

1 total trend - total # of customers - # total of transactions - total # of sales revenue 2 customers - # of customers based on the signup - 3 transactions - # of transactions over time - revenue of transactions over time - # of sales based on items - # of sales revenue based on items

This is either totally wrong or poorly posed/clarified. Firstly, as someone else pointed out WHERE is always evaluated before GROUP BY by the SQL engine. Secondly, LAG only lags the previous row. So if a property has purchases in January March and May the LAG query will return percentage changes from January to March and March to May. My interpretation of month over month delta would be Feb -> March and March-> April. So since property 9 only has purchases in non-consecutive months Jan, March and May, it wont have increasing MoM values at all (percentage change from zero to any value is infinity and from a finite number to zero is -100). Presenter should check his results to see if property 9 should be there at all. If so, he should clarify at the start what he interprets it to mean: "Calculate the percentage change in total rentals for each property for each month in 2022" - i believe his interpretation is at the least... unorthodox. 🙃

1 ctr. 2 # of keywords searched 3 whether use bundle sales

Can anyone tell me what's the proportion they are getting as a output?

As a PM aspirant, this has been such a unique way to brainstorm metrics is just beyond me. Wish I found this sooner!!

@DataInterview would cancellation rate also count as Task Success in HEART framework?

insightful!

I recently failed google ds interview in AB Testing ; Feels really bad + next attempt would be after a year which makes me even sad. So yeah if sitting for FAANG always look into AB testing and statistics

How did I not see this before the last 5 case studies I bombed 😢. Ty!! 😊

This was excellent

Nice for having a brush up! Thank you!

absolutely fantastic video! thank you so much!

Here is a simple version. You do not necessarily need a subquery or a temp table to solve this question: SELECT w.url, COUNT(a.event_id) AS click_count FROM google_search_activity a JOIN google_search_websites w ON a.website_id = w.website_id WHERE a.event_type = 'clicked' AND a.creation_dt BETWEEN '2022-01-01' AND '2022-06-01' GROUP BY w.url ORDER BY click_count DESC;

Here is a simple version. You do not necessarily need a subquery or a temp table to solve this question. SELECT w.url, COUNT(a.event_id) AS click_count FROM google_search_activity a JOIN google_search_websites w ON a.website_id = w.website_id WHERE a.event_type = 'clicked' AND a.creation_dt BETWEEN '2022-01-01' AND '2022-06-01' GROUP BY w.url ORDER BY click_count DESC;

Thank you, Dan, for consistently providing us with valuable and practical content

Is it the variance of the metric or the difference in metric between control and treatment?

can you please clarify what is one sample and two sample?

Hey Dan! This is exactly what I wanted. Not just the template but going in details of each step and explaining why we chose something vs the other. This is amazing! I am going to subscribe to practice more of these.