R help - Aug 1997 - R-beta: bug report

The following seems to be a bug when I tried to use the predict function.
> forbes
      boil.pt pressure log.pressure
 [1,]   194.5    20.79     303.4472
 [2,]   194.3    20.79     303.4472
 [3,]   197.9    22.40     310.9061
 [4,]   198.4    22.67     312.1042
 [5,]   199.4    23.15     314.1995
 [6,]   199.9    23.35     315.0597
 [7,]   200.9    23.89     317.3460
 [8,]   201.1    23.99     317.7637
 [9,]   201.4    24.02     317.8887
[10,]   201.3    24.01     317.8470
[11,]   203.6    25.14     322.4460
[12,]   204.6    26.57     327.9783
[13,]   209.5    28.49     334.9553
[14,]   208.6    27.76     332.3596
[15,]   210.7    29.04     336.8674
[16,]   211.9    29.88     339.7189
[17,]   212.2    30.06     340.3195
> lm3                               
Call:
lm(formula = log.pressure ~ boil.pt, subset = c(-12))

Coefficients:
(Intercept)     boil.pt
 -95.176624    2.051855  
> forbes.pred
    boil.pt
190     190
200     200  
> predict(lm3,newdata=forbes.pred)
Error in if (object$intercept) X <- cbind(rep(1, NROW(X)), X) : missing value
wh
ere logical needed   


For this problem, I have fixed the predict function in src/library/base/funs as
follows:
And the predict function now seems to work ok.

predict <- function(fit,...) UseMethod("predict")

predict.default <- function (object, ...) {
                 namelist <- list(...)
                 names(namelist) <- substitute(list(...))[-1]
                 m <- length(namelist)
                 X <- as.matrix(namelist[[1]])
                 if (m > 1) 
                   for (i in (2:m)) X <- cbind(X, namelist[[i]])
                 if (attr(object$terms,'intercept')) 
                   X <- cbind(rep(1, NROW(X)), X)
                 k <- NCOL(X)
                 if (length(object$coef) != k) 
                   stop("Wrong number of predictors")
                 predictor <- X %*% object$coef
                 ip <- real(NROW(X))
		 sum1_summary(object)
		 covmat_sum1$cov.unscaled*sum1$sigma^2
                 for (i in (1:NROW(X))) ip[i] <- sum(X[i, ] * 
                       (covmat %*% X[i, ]))
                 stderr1 <- sqrt(ip)
                 stderr2 <- sqrt(sum1$sigma^2 + ip)
                 tt <- qt(0.975, object$df)
                 conf.l <- predictor - tt * stderr1
                 conf.u <- predictor + tt * stderr1
                 pred.l <- predictor - tt * stderr2
                 pred.u <- predictor + tt * stderr2
                 z <- cbind(predictor, conf.l, conf.u, pred.l, pred.u)
                 rownames(z) <- paste("P", 1:NROW(X), sep =
"")
                 colnames(z) <- c("Predicted", "Conf
lower", "Conf upper",
                                  "Pred lower", "Pred
upper")
                 z
}
******************

Another problem I have is that when I use the sink function to divert output
to an external file, I used the command options(echo=T) so that the commands
will be echoed in the external file. However, the echo command doesn't seem
to work. Any suggestions?

Thanks.

Kung-Sik

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-r-help
mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Kung-Sik Chan <kchan at stat.uiowa.edu> writes:

> > predict(lm3,newdata=forbes.pred)
> Error in if (object$intercept) X <- cbind(rep(1, NROW(X)), X) : missing
value wh
> ere logical needed   
> 
...
> 
> For this problem, I have fixed the predict function in
src/library/base/funs as follows:
> And the predict function now seems to work ok.
> 
> predict <- function(fit,...) UseMethod("predict")
> 
> predict.default <- function (object, ...) {
>                  namelist <- list(...)
...

Yup. The current predict() has little relevance to what it should do.
It seens to come from an ancient version where almost everything
involving fit objects was different. A proper fix is somewhat more
tricky and should probably work like the one in Splus. Actually I'd
want it to do better because the factor-matching is such a pain
(prediction data may have different levels represented). I've been
working on it, but I got distracted by other matters. Will try to get
back to it soon.
> Another problem I have is that when I use the sink function to divert
output
> to an external file, I used the command options(echo=T) so that the
commands
> will be echoed in the external file. However, the echo command doesn't
seem
> to work. Any suggestions?
Hmm:
> options()$echo
NULL

So it isn't implemented (yet?). One option is to do 

'R < infile > outfile'

Another is to go and implement the echo option.... ;)

Or pick up the modified S-mode and run things from emacs. (It's all
in the FAQ)

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-r-help
mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=