[R] accessing confidence interval values from the 'predict' function

[Michael Dewey]

Dear Avril

I think you will find that predict.lm returns a matrix not a data frame. 
I find str() useful when R does things I did not expect or quite understand.

Michael

On 10/08/2015 09:20, alc wrote:
>
>
> Dear all,
>
> I'm wondering how can I access the confidence interval values ('upr' and
> 'lwr' values) produced by the 'predict' function. For example, I fitted
> a linear regression line using:
>
>   fit <- lm(y ~ x)
>
> I then wanted to calculate a 95% confidence interval for the line, and
> did this using:
>
>   ci <- predict(fit, data.frame(x), interval="confidence")
>
> I see that the 'ci' object has 'upr' and 'lwr' variables stored in it,
> but am not sure how to access them properly.
>
> I find I can do it using:
>
>   upper <- as.data.frame(ci)$upr
>   lower <- as.data.frame(ci)$lwr
>
> However, I'm wondering is there a more 'proper' way to do it in R? I
> find that things like ci$upr don't seem to work, but am not sure what's
> the right way?
>
> Kind Regards,
>
> Avril
>
>
>
>

-- 
Michael
http://www.dewey.myzen.co.uk/home.html