Bias-Variance Tradeoff : Data Science Basics

[@mathieutorchia178]

The value that you put out for free is INCREDIBLY appreciated. You are seriously helping so many students and professionals through your videos. Thank you so much on behalf of all of us.

[@thefuzedout]

I don't think I've ever commented on an educational video in my 22 years cause I'm always left with a doubt at the end, but this video genuinely helped me understand the terms in detail! Thank you ritvik

[@rudeprover]

As always Ritvik never disappoints when it comes to breaking down a concept without relying on mathematical equations, and still giving the best overview of a concept in the most generalized way possible. Thank you!

[@thepresistence5935]

This is the only video, I learnt about bias-variance tradeoff

[@souravkundu1735]

i have been searching for this topic all over the internet... but i have always got very complex explanations... youyr explanation finally saved the day. Thanks !

[@jannatulferdaush377]

The best video on internet about Bias Variance. Since I am interested in machine learning, I have watched hundreds of video on this topic, but haven't understood much. But your video made it easy for me. You saved my life. Thanks a lot man. I pray for your good health & wealth

[@shubhamsharma-ne2ke]

"Obviously", this is the best explanation of bias and variance.

[@liamtarr]

I really enjoyed this explanation. This is a must watch for anyone who wants to start working with machine learning.

[@ritvikmath]

Glad you enjoyed it!

[@telugujack5562]

1200 long page book on Practical and real world scenario based book on data science and machine learning. Download sample pages from the below product page. #Datascience #machinelearning #python #interview #interviewquestions Check out my latest product: DataScienceMachineLearningQuestionBank A well curated and catalogued practical based data science and machine learning questions and solutions.A must for all those aspiring and practising data scientists and ML engineers to deepen your... payhip.com

[@nicowww89]

I think you nailed it by clearly showing how model prediction vary based on the training data. Most bias-variance explanations out there never really make it clear that you're looking at the same model trained on different data Great job!

[@stanlukash33]

And here I am - coming back to your videos even after finishing an ML course. Thank you

[@subhankarghosh1233]

You are AWESOME... The fun part is, I grasped this concept earlier from some text but when I was reading some other resources about some other topics they also brought up the bias variance. From those explanations, I got completely confused and started to doubt my understanding itself. Thanks for your effort. It really helped me bring back my confidence.

[@ritvikmath]

Glad it was helpful!

[@AdityaSharma-do1ho]

Hey Ritvik, I like the way you try to build up the intuitive sense around maths rather than focusing on the theorems! Great work!

[@danspeed93]

One of the clearest explanations of this tradeoff I've seen so far, thanks!

[@DigitalOutlawed]

Man, Thanks so so so so much, you have no idea how much time you save us students.

[@honeyBadger582]

One of the best DS tutors out there. Keep it up!

[@ritvikmath]

Glad you think so!

[@mateoricucci1285]

This is the best youtube video I've ever seen. Thank you so much

[@ritvikmath]

Wow, thanks!

[@sarfrazjaved330]

Hey @ritvikmath, This man is genius. He explains the complex stuffs so simple. Hats off sir.

[@Boringpenguin]

Thank you so much. This is by far the clearest explanation I have ever come across on this topic!

[@ArunKumar-yb2jn]

A wealth of wisdom in a nugget!

[@monazaizan947]

Wow!! Very well explained!! I appreciated your effort. It is not easy to put all these together perfectly! Thank you so much!

[@HonestConsulting]

Hey bro. You make excellent videos. Keep up the good work.

[@ritvikmath]

Thank you so much 😀

[@rahulahuja1412]

The pen-toss - finger-snap combo at the end was fun. 😄 Truly appreciate this succinct summary of the concept; served as a wonderful refresher. Saving this in a revision playlist.

[@Moongchilove]

Appreciate the video. What a intuitive, straight-to-the point lecture in a perfect play time

[@aniket1983]

Thank you so much....this was one of the best explanation of variance- bias...again Thank you so much for all your videos...Respect...

[@ataulmorshed3964]

The wow part was the explanation on contribution of each model to learn the average 'signal' (true pattern) and 'noise' of a data.

[@dharmharley6871]

Pleasantly surprised to see this good of content on youtube!

[@yakovsushenok]

Simply explained is best for intuition

[@houyao2147]

very clear about the definitions of variance and bias. It tells sth. about many models, not one.

[@augustoc.romero1130]

Write a book about simply explained data science. You're great at explaining things intuitively and am sure you'll have a market for such a thing. Thanks for your vids bro.

[@kushagraagrawal7292]

FFS my university is filled with world famous research professors that dont know how to teach and couldnt explain this concept in hours of lecture :( thank you so much!

[@woowooNeedsFaith]

Subtitles of this video are rally nice. Without it I wouldn't known you speak Korean.

[@jackvaughan7265]

Made this crystal clear. Thank you for this content

[@ramiscanyakar5078]

amazing video pure informations well done thanks ritvik

[@jayneabrown]

You are an absolutely incredible teacher!

[@cara1362]

Just wanted to say thank you. All of your videos I have watched are pretty understandable. I'm reviewing those terminology concepts and prepare my coming interview. If it's possible, I'm looking forward to seeing your video about L1 and L2 explanation or the overfitting solution.

[@harrywilkinson4826]

This was a fantastic explanation. Thanks for the clarity!

[@derekfun75]

wow. You are so good! You really help me understand the concept completely!

[@ritvikmath]

I'm so glad!

[@arkophredy]

The explanation was super!!! Thanks for sharing this

[@zrinkaduvnjak8037]

Hi, I'm a pharmacist by training and at the moment doing a PhD in the field of pharmacometrics. I love your videos! It is just amazing that you are able to explain them to someone without a proper math or statistics background. In my field we are using quite a few data science concepts but in quite different (but at the same time similar) ways. I thought this video could be a great opportunity to share this with you (so you would keep explaining everything in such a simple way). In pharmacometrics, we are working with clinical trials data and developing nonlinear mixed-effects models to explain changes in concentrations of drugs in the human body over time (we have multiple drug concentration measurements over time, together with patient characteristics available). Our models are sets of ordinary differential equations and most of the parameters in these models are treated as random effects parameters (we can not pool all the samples from all the patients together, but need to account for samples belonging to the same individual, and that patients differ from one from another). Bias-variance tradeoff comes in place in the last stage of model development (after general trends were accounted for with the system of ODEs), when we are trying to find covariates on model parameters (some patient characteristics, such as age, or some blood measurement). In contrast to your field, we do not have multiple sample datasets available (splitting is not an option since the dataset size is rather small) and it usually takes years after we publish our models that somebody actually tries to validate them on external data. We opt for complex models that contain all possible covariates when we want to make inferences on covariate effects, or more precisely - when we want to prove that dose adjustment is not needed (when we want to estimate these negligible parameters), but we opt for a parsimonious model when we want to use it for simulations (other clinical trials, same special populations such as obese people etc.). Also when we are talking about bias in oversimplified models, I believe in the first place we mean that over parameter estimates (and not predictions) are biased, since if we do not include 2 covariates, but only 1, and the two are correlated, estimate of one will also account for not including the second one (estimated effects seems to be higher than it is). After watching your video I finally understand why we are saying that simpler models are more predictive. Thank you!

[@keremaksoy8007]

Dude, u are a perfect teacher

[@mprawesh]

My course book definition was so confusing but you made it so clear. Thank You!

[@ritvikmath]

Of course!

[@tymothylim6550]

Wonderful video! Never really understood these terms when studying until now :) Thanks a lot!

[@amyliu2777]

This makes so much sense! Thank you for the awesome explanation.

[@anjamisimovic9214]

you are just GREAT dude... you explained it sooooooo nicely i cant believe

[@justin.c249]

Very well explained! Great work!

[@omarz8390]

excellent presentation for boas and variance. Thank you

[@王十八-k6w]

Best explanation on this topic! Thank you!

[@navanarun]

Thanks for simplifying such a complex topic!

[@DrNoureddinSadawi]

Great explanation Ritvik, thank you!

[@GeoffryGifari]

So should we strive to optimize the _product_ of bias and variance?

[@ICEICISIC]

Few variables, consistent prediction across datasets, average prediction does not get closer to true value. Many variables, inconsistent prediction across datasets (as predictor function models the idiosyncratic noise in the data - leading to high variance in predictions or prediction decisions) , average prediction gets closer to true value (as noise from individual datasets cancel each other - leading to low bias or error between the average across predictions and the true value).

[@sameerpatel4731]

Excellent explanation - thank you for this video

[@ritvikmath]

Glad it was helpful!

[@chuckyneoable]

Awesome explanation. Sincere thanks

[@DarkShadow-tm2dk]

Bagging and Boosting ❤️ Can u please part 2 for this video with mathematics bcoz you really are good at explaining complex things I understood pca fully bcoz of u

[@andreyb7419]

Love such an easily explained complex stuff! Thx a lot!!!

[@ShakrinJahanMozumder]

Can't thank you more! You're amazing, man!

[@ritvikmath]

Happy to help!

[@ryansandan7897]

this is the best explanation!

[@AriaHarmony]

Thank you very much this's exactly what I needed.

[@mosama22]

One more GREAT video, really don't know what to say man, thank you :-)

[@estagidados302]

great explanation. thank you, master

[@r.walid2323]

Thank you, what a great explanation

[@699ashi]

Yes, I did intuitively understood the concept. Thanks

[@xavecoding]

Amazing explanation... tkx a lot

[@mango-strawberry]

wow. perfectly explained. holy moly

[@jwdex]

Thank you so much! Great explanations, wish I would've watched this sooner

[@Werdnabae]

love the marker flip at 5:39 LOL

[@zephyrsurfteam]

Well said! Keep up the good work.

[@ritvikmath]

Thanks, will do!

[@jakubjan44]

Really good explanation

[@smilingbread22]

I like how u explain it, can u make videos about LASSO, SCAD and MCP, I still confused about them..

[@ritvikmath]

I do have a LASSO video: kzbin.info/www/bejne/oJPahHatpLloZ68 And thanks for the other suggestions!

[@soajack]

Great explanation ! Thanks !

[@ariellise]

very clean explanation

[@user-or7ji5hv8y]

Recommend future video suggestion: Fisher Information.

[@arrozenescau1539]

wow you make it so easy to understand

[@ritvikmath]

Thanks!

[@alexcessy]

Great explanation

[@ritvikmath]

Glad it was helpful!

[@SPeeDKiLL45]

man so good explained really.

[@sairabaig7272]

Sir great explanation, plz make videos on statistical inference

[@hiteshthakur416]

Being a researcher, I should say that when you defined the term bias, you should also explain what is variance else this isn’t for dummies.

[@ebateru]

Great video man! Tnx a bunch!

[@ritvikmath]

Glad it helped!

[@mrbeancanman]

really great explaination thanks!

[@ChocolateMilkCultLeader]

Fantastic explaination

[@alejandrocanada2180]

what a great video, thank you very much

[@P4ExHzLRuuiFMg3X4U3v]

Great presentation! You mentioned that complex models have a tendency to pick up both real patterns and noise from the dataset they are trained on, so their predictions are all different (high variance) due to the noise, but correct on average. I'm wondering, if you have a very large dataset, why isn't it viable to train the same very complex (strongly overfitted) model on N chunks of the dataset and use the average prediction of these N models?

[@MNCAMANI15]

This man is amazing.

[@canernm]

Very good video. Thanks!

[@ritvikmath]

Glad you liked it!

[@diegososa5280]

Thank you very much, as always spot on

[@GeoffryGifari]

High variance doesn't seem to bad... is the downside the lack of efficiency (computational cost too high for overly complex models)?

[@peterc.2301]

Absolutely Perfect!!!

[@leonhardolaye-felix8811]

Amazing video

[@karimmohanad2k01]

Thanks for the effort!

[@baurks]

This is great. But what exactly is bias? I heard 2 different things, 1) it means how different the different models are in terms of their predictions, 2) how far their predictions are from their true values If I got it these right, definitely confusing. Could you please clarify? Thank you

[@SergioPerez-wj1nz]

Great explanation. I hope you are a teacher somewhere.

[@prathammehta4290]

Well Explained!!

[@yuckbutyup]

Thank you for the video, great job! But I have a question that based on my understanding, variance here should be the difference between prediction accuracy from different test data set. Please correct me if I am wrong. Thanks again for the video, love it.

[@jierunchen873]

Nice tutorial. Suddenly at around the 8th min, the message behind the bias-variance just pop out!

[@ritvikmath]

Glad it helped!

[@nidasiddiqui21]

Hello..Can you please make some videos on Statistical Copulas-- Vine copulas. There is no good explanation around.

[@janarthananpoornavel]

Thanks a lot ritvik

[@ritvikmath]

You're welcome 😊

[@gadmuhirwa5226]

does Variance same as Loss? bcz even in the formula seems the same = [sum(x-xhat)^2]/n @ all

[@lsacy8347]

how would an average of all the overtrained models look like? It would produce the right answer, does it mean the noises are somehow cancelled out?

[@ramankutty1245]

Very well explained. I love your videos, though you speak really fast!

[@ritvikmath]

Thank you! 😃

[@sakhawathossain3362]

your explanation is as easy as catching the pen😄