"Peter Wollan" <wollan@mayo.edu> wrote
> Another problem with importing from Excel: if a column is mostly numeric,
> but has a few character values somewhere in the middle, S-plus imports the
> column as numeric and replaces the characters with NA. Columns which have
> enough characters in the first few rows (100?) are imported as character.
> Columns which are mostly dates are imported as dates, but deviations from
> the date format are killed.
>
> I'd like to propose a solution: allow the option of importing from Excel
> ***entirely as character data***. This would enable the user to see what is
> actually in the file, and to convert to numeric through functions which
> compare with the original and flag deletions.
>
> The Import routine, as it is now, runs into mysterious S-plus conventions,
> which interact with mysterious Excel conventions, to produce serious data
> problems. I'd be pleased to hear better alternatives than importing as
> character data (which can't even be done now).
>
> This may only be a problem to people who use S-plus in an environment where
> Excel is used as a data recording or transfer tool, but surely there are
> more than two of us.
>
> Peter Wollan
> wollan@mayo.edu
Are you using S-PLUS4 or 4.5, or 2000?
In version 4.5 it appeared that the contents of the first row
determined the mode for the whole column. It ignores (at all events
with Excel 97) rows beyond number 16384. All this seems undocumented.
Users need to be warned that, because fo these various interactions,
the results from any import from Excel must be checked with unusual
care.
Does anyone know whether the behaviour has changed in S-PLUS2000?
John Maindonald email : john.maindonald@anu.edu.au
Statistical Consulting Unit, phone : (6249)3998
c/o CMA, SMS, fax : (6249)5549
John Dedman Mathematical Sciences Building
Australian National University
Canberra ACT 0200
Australia
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|