The weight is unbalaced. So does they heavier side land face down?

An excellent video

Great primer

Thank you very much! 🔥

So bad

are you an angel??

Super helpful info. Thank you!! Where can we see the complete pie charts for ML and stats? Is there also one for Python coding? I've been searching in your channel and didn't have the luck to find them.

*Usually* not but "never" is too strong. Layoffs also target based on compensation, age, business strategy changes -- these are independent of performance. If you are top 10% and your company (e.g. a large bank) shifts away from your business or closes your division it doesn't matter. These things also happen to top 10% employees during M&A.

Is dsa asked in data acience interviews?..as i understand a data scientists job is to find insights from the data and build models in python using pandas library..

I think you should have choosen a different example to explain metrics to select other than the youtube one, it would have been more beneficial to understand.

❤❤❤❤

Great work, Emma. Appreciate your efforts.

2:11 4:50 7:55

Excellent video Many thanks. Could you kindly make a video for time to event with survival SVM, RSF, or XGBLC?

amazing thank you!

Good explanation

Great tutorial! Thanks for sharing this valuable content.

Good explanation but I think the last error is incorrectly handled.... Imagine you run an experiment and it is significant (you haven't checked the observed power yet), if you accept it then it is wrong but if you rerun it you just nearly doubled the p value. We should be only looking at the rerun or let the experiment have significant power (probably more than our threshold)

1:41 "It should not be obscure like what you see in Wikipedia" 😅😁😁

wow it's very informative emma

I think the green and blue parameters are swapped for the normal distribution diagram.

Really bad video. Keeps iterating the same thing.

Can you explain how this query works ? SELECT COUNT(p) OVER(order by p ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) as p_count, p, c,d FROM `d.t` SELECT COUNT(p) OVER(order by p ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) as p_count, p FROM `d.t` Why is the window affected when I specify c and d columns? I would expect to count 1 before, current one and next one. So for first row it would be 2 for next ones except last one 3 and for last one to be 3, because the window is made of 3 items so it will count 3 most of the time. But no.. it returns different counts that does not make any sense to me. I read the bigquery documentation and I really don't get it.

1 , nice to meet you

I wish all professors explain it just like you do, well done!!

This video helped me clear an interview. Subscribed. Thank you.

thanks a lot!!

Thanks Emma and Yuan! very useful videos!

fantastic lecture! thanks Yuan and Emma!

Emma your videos are amazing! Keep doing what you do. Looking forward to more such content

can we contact you for consultation?

Excelent recommendations !

Okay subscribed !

Such a poor pronunciation

Hi Emma, I have a doubt - How would one calculate the average time spend per user per day? Say we select a random sample of 10 users as in your example in the video. For those 10 users we have data on the time spent per day for each of the users. Now a user might have multiple time spent per day values depending on if they were active on several days. So for a particular user we calculate the average time spent per day by that user and then take the average time spent for the 10 users using average of individual averages?

I'm a generic SWE who spends a good amount of time in data and wants to understand machine learning better. This has been wonderful, thank you so much for putting it together.

Stop producing new videos?

Hi Emma, how to access the links to LinkedIn profiles in Action Items?

我初中时候班里男同学最喜欢的就是一个叫emma的女同学，成绩超好，长得也美😏

Thanks for your recommendation, Emma. I was looking for a book on A/B testing and I was fortunate that I saw this video.

Thanks, great video - clear explanation ☺

Thank you

Super helpful, thanks Emma!

This is a great overview of methods of feature selection. This is just what I needed! Thank you

Very useful. really appreciate your effort in giving the educational session to the public!!!

Some details who'd like to to get into that really deep: regarding L1 penalty - we can't really choose w2 = 0 and w1 = 1, since loss consists of 2 parts: loss from the diamond shape (l1 penalty) + loss from the ellips shape (initial loss function). since for (w1, w2) pairs (0, 1) and (1, 0) l1-penalty term values are the same and equals 1 + 0 = 0 + 1 = 1, we now look at the initial loss function as it just as well depends on the (w1, w2) values. for w1 = 0 and w2 = 1 (closer to the center point) loss is going to be less than for w1 = 1 and w2 = 0 (farther from the center point) - we can see that from the contour lines plot. therefore, optimizer won't go there and will converge on w1 = 0 and w2 = 1 and that's it.

Video was made in 2021, and I busted into laughter when “company X” is mentioned 😂 who would know, right? 😂😂😂

Creating a personal brand is the best way to compete in this market, when u become an expert, u don’t care about crisis

The table at ~9:00 is wrong. should go with running total, running average, etc.

where I can download this template ?