R devel - Nov 2008 - Problems with packages fda and splines (PR#13263)

Full_Name: David D Degras
Version: 2.8.0
OS: Mac OS X 
Submission from: (NULL) (128.135.239.11)


I have recently installed the version 2.8.0 of R along with package fda (v
2.0.2)
and its dependencies (including package splines v. 2.8.0).

Here are my problems:

1) The package splines should feature functions such a predict.bs,
predict.bSpline and such and it does not! I can make calls to bs, ns, and
interpSpline but not to any predicting function. 
Example:
> library(splines)
> predict.bs
Erreur : objet "predict.bs" non trouv? # in english: object
"predict.bs" not
found

Also, I cannot track the package splines on the CRAN website. Why is that?

2) Package fda: it does not handle at all NAs although promising to. 
I have met this problem using Data2fd and project.basis as well.
An example script is enclosed at the end of this email.

Can you help on this ?

Thanks,

David

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
R SCRIPT for Data2fd with NAs 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> library(fda)
> y=runif(10)
> x=Data2fd(1:10,y) # creates functional object in B-spline basis
Warning message:
In smooth.basis(argvals, y, fdP, wtvec = w, fdnames = fdnames) :
  The number of basis functions = 12 exceeds 10 = the number of points to be
smoothed.  With no smoothing (lambda = 0), this will produce a perfect fit to
data that typically has wild excursions between data
points.
> x$coefs # here it works because no NAs
            [,1]
 [1,]  0.9193563
 [2,]  0.1590327
 [3,]  0.6441482
 [4,]  1.1257380
 [5,]  0.1234022
 [6,]  0.7228168
 [7,]  0.8709656
 [8,] -0.3634714
 [9,]  0.9097221
[10,]  1.1726878
[11,]  0.4130424
[12,]  0.3122870
> y[3]=NA # create a NA
> x=Data2fd(1:10,y)
Warning message:
In smooth.basis(argvals, y, fdP, wtvec = w, fdnames = fdnames) :
  The number of basis functions = 12 exceeds 10 = the number of points to be
smoothed.  With no smoothing (lambda = 0), this will produce a perfect fit to
data that typically has wild excursions between data
points.
> x$coefs # vector of NAs because y[3]=NA
      [,1]
 [1,]   NA
 [2,]   NA
 [3,]   NA
 [4,]   NA
 [5,]   NA
 [6,]   NA
 [7,]   NA
 [8,]   NA
 [9,]   NA
[10,]   NA
[11,]   NA
[12,]   NA

Hello, David: 

      Thanks for the comments about NA problems in 'fda'.  I'll look
into this and get back with you. 

      Spencer

degras at uchicago.edu wrote:
> Full_Name: David D Degras
> Version: 2.8.0
> OS: Mac OS X 
> Submission from: (NULL) (128.135.239.11)
>
>
> I have recently installed the version 2.8.0 of R along with package fda (v
> 2.0.2)
> and its dependencies (including package splines v. 2.8.0).
>
> Here are my problems:
>
> 1) The package splines should feature functions such a predict.bs,
> predict.bSpline and such and it does not! I can make calls to bs, ns, and
> interpSpline but not to any predicting function. 
> Example:
>   
>> library(splines)
>> predict.bs
>>     
> Erreur : objet "predict.bs" non trouv? # in english: object
"predict.bs" not
> found
>
> Also, I cannot track the package splines on the CRAN website. Why is that?
>
> 2) Package fda: it does not handle at all NAs although promising to. 
> I have met this problem using Data2fd and project.basis as well.
> An example script is enclosed at the end of this email.
>
> Can you help on this ?
>
> Thanks,
>
> David
>
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> R SCRIPT for Data2fd with NAs 
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
>   
>> library(fda)
>> y=runif(10)
>> x=Data2fd(1:10,y) # creates functional object in B-spline basis
>>     
> Warning message:
> In smooth.basis(argvals, y, fdP, wtvec = w, fdnames = fdnames) :
>   The number of basis functions = 12 exceeds 10 = the number of points to
be
> smoothed.  With no smoothing (lambda = 0), this will produce a perfect fit
to
> data that typically has wild excursions between data points.
>   
>> x$coefs # here it works because no NAs
>>     
>             [,1]
>  [1,]  0.9193563
>  [2,]  0.1590327
>  [3,]  0.6441482
>  [4,]  1.1257380
>  [5,]  0.1234022
>  [6,]  0.7228168
>  [7,]  0.8709656
>  [8,] -0.3634714
>  [9,]  0.9097221
> [10,]  1.1726878
> [11,]  0.4130424
> [12,]  0.3122870
>   
>> y[3]=NA # create a NA
>> x=Data2fd(1:10,y)
>>     
> Warning message:
> In smooth.basis(argvals, y, fdP, wtvec = w, fdnames = fdnames) :
>   The number of basis functions = 12 exceeds 10 = the number of points to
be
> smoothed.  With no smoothing (lambda = 0), this will produce a perfect fit
to
> data that typically has wild excursions between data points.
>   
>> x$coefs # vector of NAs because y[3]=NA
>>     
>       [,1]
>  [1,]   NA
>  [2,]   NA
>  [3,]   NA
>  [4,]   NA
>  [5,]   NA
>  [6,]   NA
>  [7,]   NA
>  [8,]   NA
>  [9,]   NA
> [10,]   NA
> [11,]   NA
> [12,]   NA
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>