Regression with Count Data: Poisson and Negative Binomial

Poisson, quasi-Poisson, and negative binomial regression - when to do them and how you should choose the method. What are overdispersion and underdispersion, and why are they problems? How to deal with too many zero counts (zero-inflation) or when zero counts are impossible (zero-truncation).
0:00 Background
2:26 Poisson Regression: What and Why
7:05 Overdispersion: Quasi-Poisson or Negative Binomial
13:25 Zero-Inflation and Zero-Truncation
18:37 Summary Table

[@youngzproduction7498]

This clip alone gave me more information than I ever imagined. Thanks.

[@Sefa-Safa]

This is super useful to understand and apply the poisson regression analysis. one of the best tutorial I have ever watched. Many thanks!

[@cathrineb4923]

Thank you so much for posting this video!

[@hola-kx1gn]

Nice ! Thanks for the timestamps !

[@RqueErre]

Amazing video! Excellent explanation and very useful!

[@laurafigueroa44]

super helpful, thank you Matthew!

[@pranilbasu7476]

Great ! Please keep posting more such videos

[@dtox316]

Best video on the internet on this topic in my opinion. Covers the why, the what, the when and the how.. perfect! Liked and subbed. Thanks!

[@smartmarc333]

I agree, really really good educational video, much appreciated

[@shengmingwong6968]

Very excellent video! Thanks!

[@pabos1993]

Congratulations! Excellent Video. It would help a lot if you provided the code for the variance vs mean plot and the corresponding lines predicted by the quasi-Poisson and negative binomial model. Thanks

[@sameerhassan7516]

Very nice video. Thank you!

[@HashanDananjaya]

Thank you very much! This helped me quite a lot!!

[@kanefoster8780]

great vid, cheers

[@estefaniavillanueva1294]

OMG, thank you so much for this very informative video, it really helped me a lot!

[@francoisdaudelin8610]

Thank you so much! Excellent explanation!

[@francoisdaudelin8610]

Quick question: Can I use multivariate zero truncated Poisson regression to compare the effects of two IVs on a DV? I'm mostly wondering if the size of the coefficients lose their meaning because of the zero truncation

[@steilacoom]

Great explanations

[@SangheiliG099]

Just thank youi!

[@laxmanbisht2638]

great.. thanks sir

[@iaaan1245]

this is an awesome video

[@ibrahimkassoumhabibou4076]

Hi ! Thank you for explaining theses models. Is it possible to provide the R code you 've been using to compute the graphics ?

[@kaustavchakrabarty4925]

can you provide the code for the mean vs varaince plot for quasi poisson and negative binomial glm....that will be very helpful

[@Nicoleuni7]

THANK YOUUUUUU

[@tonycardinal413]

Excellent video. Thank you ! Quick ques: When you say that the mean must equal the variance, do you mean that the mean of all the Y values of the observed data points must equal the variance of all the observed Y values of the dots of the scatter plot? thanks !

[@abcpsc]

So... How to get the confidence interval of y after fitting the model?

[@brazilfootball]

Do fish counts from a bay over time disqualify this kind of data for a Poisson distribution because it's a time series dataset?

[@thejll]

When you talk about y being Poisson distributed, do you mean ‘the errors on y’? I have data where y is a combination of things only some of which carry counting-statistics uncertainties.

[@BrOgam3rHD]

Holy fuck is this video good