Video není dostupné.
Omlouváme se.
POLS 506: Bayesian and Nonparametric Statistics - Lecture 10 - Missing Data and Multiple Imputation
Vložit
- čas přidán 30. 11. 2012
- Created on 12/1/2012 by Dr. Justin Esarey, Assistant Professor of Political Science at Rice University. Notes problems that can arise from ignoring missing data in statistical analysis. Discusses two potential solutions: the Multiple Imputation using Chained Equations algorithm of van Buuren et al., and a fully Bayesian model of imputation in BUGS.
@Justin I appreciate I am 8 years too late for this to matter but the issue at 1:28:28 is that when you use as.numeric() on a factor it converts it into 1 for the first level and 2 for the second level (normally you would use as.numeric(factor) - 1 to get the proper binary representation). As you have no intercept in your model (and due to R using corner point constraint by default) the first factor becomes your intercept (this why your estimate for the intercept is ~1) and the coefficient for z2 is the offset from the intercept which is 2-1 = 1 which is why your z2 estimate is also ~1. At the very least I hope this helps anyone else who watches this video in the future and is confused :) Otherwise thank you for such great content this has been super useful !
No one links better the theory of handling missing data and the practical approach! Thanks a lot! Hope to see further videos. SR, Germany.
Thank you so much. Your lecture is awesome!! I've never seen this (relatively) easy to understand lecture before.
Excellent lecture! I hope you post more of these online.
great lecture! Thanks!
Thank you very much. I really liked it. Was very useful for my study
This is very helpful. Thanks
Thanks. This is really helpful. Though I have a doubt about the formula appears at 1:00:42. It seems not correct?
Great video!TKS! would u please provide Lecture 7,Lecture 8,Lecture 9?
Is the R code that is shown in the lectures available?
1:06:57 kkkk funny
mice is nice kkk