[R] glm package - Negative binomial regression model - Error

Paula Couto paulaveronica at gmail.com
Mon Feb 26 02:07:24 CET 2018

HI there

I am running this model in negative binomial regression, using glm.
I had no problems with running the model with a set of data, but now that
i'm trying to run if for new one.  I always have this same error when
running the regression:

> #Run Regression
>     x=cbind(factor2ind(d$year),factor2ind(d$month_week))
>     out<- glm(cbind(influenza, n_sample) ~ x, family=quasibinomial,
> data=d)
>     d$prop<-out$fitted.values

Error in `$<-.data.frame`(`*tmp*`, prop, value = c(0.0486530542835839,  :
  replacement has 208 rows, data has 365

>     d$n_p1<-d$prop*d$factor*10
>     obs<-aggregate(d$prop, by = list(d$month_week), FUN=summary)
>     pred<-aggregate(d$n_p1, by = list(d$month_week), FUN=summary)

By the way, I previously prepared the data set  and defined that:
>         d$n_sample<-(d$n_muestras*d$factor*10)
>         d$prop<-(d$influenza/d$n_sample)

But I still don't understand why it keeps saying that dataframe has less
replacements than rows.
Could anybody help me with this?
 Many thankss!!!

	[[alternative HTML version deleted]]

More information about the R-help mailing list