What is A/B Testing? | Data Science in Minutes

What is A/B testing? In this quick tutorial, we go over the basics of A/B testing, as well as answer some in-depth questions such as: why should businesses conduct A/B testing? Or how do you perform an A/B test?

A/B testing aims to determine not only which technique performs better but also to understand whether the difference is statistically significant.

Table of Contents:
0:00 Introduction
1:29 Sucess criteria
2:58 What happens if...
3:15 Conclusion

--

--

Unleash your data science potential for FREE! Dive into our tutorials, events & courses today!

--

📱 Social media links

--

Also, join our communities:

_

#datascience #abtesting

[vijayswearingenutube]

I'm a data science student and found this brief video informative, especially if you land a job in marketing research, etc. We didn't cover this in school and I wondered why that was because I see it everywhere on job searches. What specific tools are used for this?

[kingofthedong]

im a graphic designer with no idea what i am doing other than design beautifully, and i was tasked on learning how to do a/b testing for my first job. this is very helpful for me to understand the basic, thank you !

[analisamelojete]

Basically, this is hypothesis testing but with a fancy name. Things like means difference, T tests, F tests, etc. Just pure statistical inference.

[zxvbhvxz]

This was super clear, thank you so much

[Gibson-xnxk]

Thank you so much. Extremely helpful information!)

[canaykoseoglu]

Hello, this is extremely useful information. I'm running AB tests with WASK. I can run A/B tests between my Facebook and Instagram ads and my Google ads. Furthermore, based on the results of my AB tests, I can disable inefficient ads and continue with my efficient ads.

[jarombray]

Super helpful! I loved the visualization. It helped me internalize the information.

[vf]

i fell in love with A/B Testing within a few seconds watching this.

[AL-pjjg]

good job. simple and easy way of explaining.

[TheIronhideChronicles]

Thank you very much for explaining this in simple terms

[brajeshprajapat]

thanks to give the information on a/b testing it helped me alot.

[hemantsharma]

How can i find web log dataset on healthcare website for a/b testing

[NMLeepsa]

I am curious to know how you had that background with graphs. anybody who can help? I want to make my own videos for my channel with such visuals in the background. Is there any software for that? any youtube video to explain how such video is edited

[rajgurrala]

The presenter is Amazing. But I felt she is a bit nervous, But your nervousness is pretty attractive. Come on. you are the best and you will look great being more confident. Very informative 3mins.

[VVV-wxui]

Am still not clear on A/B Test in DataScience. While I understand that this video is made to educate folks, but the message and the education part seem missing as I did not get anything from this video. Probably, the host is reading it out from the Tele-prompter and did not explain the need, approach and the conclusion. @"Data Science Dojo" -- pls pay attention and post right videos.

[whatisahandle]

Question to users: For HR/benefits (Eg health insurance) and work intranet tools, is it just me, or are websites getting more and more clunky, click & menu happy, cumbersome, and poor in their search capabilities?

All the new web tools my (large) company or new benefits groups have put out in the last ~3 years have seemed more and more like a Lego video game: you’ve got to click all the text bc the hyperlinks aren’t obvious, look carefully and explore all the text because of fancy picture & explanations vs simple titles or obvious organization, scroll way down on ever page, and explore all the hidden menus, because nothing is simply presented, and everything requires multiple clicks, even common actions.

[somanshusharma]

The fact that its required item on job posting is absurd! If you can't trust your candidates to learn this then you should not be hiring them in the first place

[davidfitcher]

why don't we just call it Version Testing?

[ZanetDesign]

Hi
Here is a question that I just can't find an answer to regarding thumbnail split testing.
when is the best time to split test thumbnails?
I use Tubebuddy for split testing my thumbnails right from day one. But im worried im doing something wrong and getting unreliable results. here is my process...
1. I design 2 thumbnails that I think will work.
2. I release the video to the public with my thumbnail A
3. I now splittest to thumbnail B and rotate daily
4. I let it run until a winner produces a statistical significance

here is the 2 questions no one seems to consider or answer

A)
Isn't the first few days an unrealistic test as the first 24 hours the CTR is not relevent to the thumbnail but is more due to alerts and subscribers seeing you have a new video out. This is almost impossible fairness as a split test goes, so my gut feeling is don't split test in the first week to avoid this.

B)
the fact that a thumbnail has potentially 2 very different audiences also makes split testing irrelevant. if my video is for search, the audience AND CTR is totally different to the suggested video audience.

Question
how do you deal with A and B?
so what and how is this meant to be handled, what do you do to allow for and address both these realistic scientific approaches to CTR?

any response would be appreciated
Your biggest fan
Zane
Zanet Design

[vivekjoshi]

Sorry but really not clear. I watched this twice & still not clear. Can anybody explain me ?