Multicollinearity

Рет қаралды 202,487

Күн бұрын

This video explains what the issue of multicollinear regressors causes for estimation, using the example of TV and Radio advertising. Check out ben-lambert.co... for course materials, and information regarding updates on each of the courses. Quite excitingly (for me at least), I am about to publish a whole series of new videos on Bayesian statistics on youtube. See here for information: ben-lambert.co... Accompanying this series, there will be a book: www.amazon.co....

Пікірлер: 50

@mreighthamburger7485 9 жыл бұрын

oh man you should be my professor you know what watching your videos helps a lot more than my professor's lectures....

@_Anonymous_9 7 жыл бұрын

Ben is Da Bomb - made it from 1-60 videos so far, actually quite enjoy studying econometrics now xD Cheers Ben!

@nonah133 4 жыл бұрын

Saving me right now with online classes. Thank you!

@rafkatakhmetkaliyev879 4 жыл бұрын

me too

@tunahanuzun603 6 ай бұрын

This video is crazily good! Never understood econometrics better, and it's actually making fun to study it! :)

@pushypin 11 жыл бұрын

Excellent presentation. I'm watching your videos to better understand the quant section of CFA Level II. Thank you Ben!

@SpartacanUsuals 11 жыл бұрын

Hi, thanks for your message, and kind words. Best of luck with the CFA! Hope it all goes well. Cheers, Ben

@LSchork17 Жыл бұрын

9 years later I am doing the same thing. Hope they were helpful for you. Thanks Ben!

@My_Melody_Forever 6 жыл бұрын

You have genius teaching skill.

@spacedustpi 6 жыл бұрын

Thanks for this explanation. So my understanding is that multicollinearity is only worth finding out if you want to know how much each attribute is contributing to the model. Which, if you want to be prudent, you should find out. So how would you find out? Run a regression twice where with one of the attributes held out in the first regression and the other held out in the second regression? Then compare the two results to determine which one has more effect on the sales? Thank you in advance.

@Tinasheziki-f5d Жыл бұрын

Thank you doctor for the presentation especially exemplification

@dandorsano9982 7 жыл бұрын

You're awesome Ben! Very helpful videos

@TJAcademyofficial 4 жыл бұрын

For more helpful videos on the subject, Subscribe TJ Academy kzbin.info/door/Q7Cbm57341QKdgZ_fTDGvw For Multicollinearity English (with EViews): kzbin.info/www/bejne/fqC3aGt9eL94rtE Urdu/Hindi: kzbin.info/www/bejne/gYbXcmmQrN6bprM (Headphone recommended for this video only)

@AbdulAziz-gt8oo 4 жыл бұрын

Thank you! its so helpful, the explanation is easy to be understand.

@TJAcademyofficial 4 жыл бұрын

@danielr3544 5 жыл бұрын

Youre Videos are great short but well explained !

@TJAcademyofficial 4 жыл бұрын

@seanmenzies1986 8 жыл бұрын

Very well explained and demonstrated. Many thanks.

@moobadaa 10 жыл бұрын

Hi Ben. I am trying to watch the videos based on the order in the playlist. But you've not talked about R2 and significance level yet and now are using these concepts!

@SpartacanUsuals 10 жыл бұрын

Hi, thanks for your suggestion. I realise there are some conflicts here and there. I will add a link to these topics in the video. Best, Ben

@samjab9237 Жыл бұрын

you're a king mate

@amitrv007 5 жыл бұрын

Hi Ben, which software do you use for these illustrations?

@chh376 8 жыл бұрын

Hi, Ben, it's really helpful. But I was wondering if we need to check the multicollinearity for variables like dummies and time trend. Because I suppose for example dummies for structural break should be highly correlated to some variables and that is the point of using them, right? and same for time effect.

@SpartacanUsuals 8 жыл бұрын

+CH H Hi, thanks for your comment. Yes, it is possible for multicollinearity to occur with dummies and time trends. Imagine that you have time series data on the sales of ice creams. In the summer there will be higher sales than the winter. You could either model this using a variable like temperature, or indirectly using a dummy which is 1 when it is summer, and 0 otherwise. These variables will be highly collinear, because they are both attempting to measure the same thing. Intuitively regression is going to find it hard to differentiate between the effects of the dummy vs the temperature variable, and hence collinearity may be a problem here. Hope that helps! Bests, Ben

@OutperformMP 4 жыл бұрын

Can someone explain why the standard errors for the B-coefficients are getting bigger because of the multicollinearity?

@atrijitdas1704 4 жыл бұрын

have a look at the formula derived for standard error of estimates of coefficients in whatever book you are using. You'll find that it has a term involving correlation of the independent variables. A high correlation, that is Multicollinearity, hence inflates the standard error

@spikeymikey9494 8 жыл бұрын

Thanks for answering my previous question. I was wondering if you could answer my question which is related to multicollinearity. The question gives you 4 auxillary regressions. One of them is... logX1 (t ratios) 0.96 (2.56) -0.83logX2 (3.49) 0.95logX3 (5.66) 0.6logX4 (3.79). I presume the parentheses are standard errors. But how do you perform an f test on that to confirm multicollinearity (related to previous part of the question).? Your help would be greatly appreciated!

@MeghanaHM 5 жыл бұрын

Why does this occur only in regression problems and not in classification ?

@Ben2020able 10 жыл бұрын

thanks man why do you think when we make centering for our variables the collinearity disappear?

@SpartacanUsuals 10 жыл бұрын

Hi, thanks for your message. Making variables standardised can reduce correlation between the two estimators in question. However, it does not remove collinearity in general between two variables. Hope that helps! Best, Ben

@siddharthadas86 7 жыл бұрын

Hi Ben your explanations are really good. Do you have any videos on multilevel or hierarchial modelling explaining the math of it?

@shreyaschaturvedi1933 4 жыл бұрын

Brilliant, thank you!

@TJAcademyofficial 4 жыл бұрын

@tmuffly1 5 жыл бұрын

Dr. Lambert, I really enjoy your videos. I have two continuous variables: rcs(Age, 5) and rcs(GRE_score, 6) that I relaxed the cubic splines on and now I a getting huge VIF values for each of those variables. Does VIF work with variables that have relaxed cubic splines please? Thank you for your important work.

@alteredkill6109 7 ай бұрын

I think if you try to Reinfeld equation against the null thatll help. Wald's theory of VIF works with variables that have relaxed cubic splines. Remember: your GE number might be low when using the SRI technique on VIF. Good luck!