How do you figure out the *ideal* sample size ... as a data scientist?

Too few samples is a problem. And ... too many samples is a problem. How do we figure out just how many we need?
Link to Code : github.com/ritvikmath/KZbin...
My Patreon : www.patreon.com/user?u=49277905

[@abdnahid]

Hey man, your channel is probably the best if not one of the best channels on youtube. You should write a book on first to last data science concepts with code and explanations! The way you teach things in such a simple manner is absolutely amazing. I understand all the hard concepts thanks to your intuitive explanation. Thanks!

[@ritvikmath]

Thanks!

[@ResilientFighter]

@@ritvikmath u should. Love your passion for stats. Really inspirational

[@WatermelonSaurus]

the way you present all these explanation is so easy and clear to understand, which could have been hundred times fuzzier had it been done from "usual professor".

[@adlee0705]

As always, so good! I always love how you explain concept without using big words. Thanks for making it simple to understand😊

[@ritvikmath]

Thanks!

[@teddy5474]

Appreciate your video man! Keep up the good work!

[@fktx3507]

Very well done. I've been doing stats for roughly 10 years but learned something new today. Thanks.

[@dl8310]

Loved the video xD!!! The code also helped a lot in understanding the concepts (CS Msc student here). Please keep doing a great job.

[@pipertripp]

This looks great!! Thx for sharing.

[@alirezamogharabi8733]

Great explanation 🙏🙏

[@TobiT822]

In physics there is the rule of thumb, that a difference in mean values is probably significant if that difference is larger than three standard deviations (3 \sigma) of the distribution. As the standard deviation shrinks with the square-root of the sample-size (the central limit theorem), one can easily estimate the number of required samples. From this one also gets an intuition why we need more samples to have a lower standard-deviation to significantly confirm a smaller difference in the mean of both distributions.

[@elhamkarami9919]

Great explanation

[@djangoworldwide7925]

Great vid, that could be an intro to "why you should always consider effect sizes and not just CI/p values"

[@afandidzakaria6881]

I really admire your channel. I am still waiting for your video to explain Kriging mathematic.

[@ritvikmath]

Thanks for the suggestion!

[@pipertripp]

Really cool. I’m going to have to look over the python.

[@pgbpro20]

Well, that was easy!

[@prasunkumar2106]

True Gems ❤️❤️

[@ritvikmath]

Thanks!

[@hirok6649]

bruhhh you triggered ma trypophobia

[@SimoneIovane]

I didn't follow the whole video through the end, but doesn't it boil down to the type I and type II errors you establish before hand?

[@nirshahar9884]

Or... you can directly derive a formula using the Chernoff-Hoeffding inequality