Odd behavior of a function within apply

Yes, David, the function described seems to insist it be of type integer
or type character and if the type was double or others might well fail as y
would never be initialized.

The goal seems to be to count how many "missing" values are found as in NA
if a numeric type or an empty string if character.

But you can have some form of NA in all kinds of object types including
character as in this construct:

x <- c("a", NA, "", "b", "NA)")
x

[1] "a"   NA    ""    "b"   "NA)"

The above has three useless elements if both NA and "" are considered
empty. So logically the condition could be to count NA and IF it is of type
character, also count "".

So rather than play games testing not just is.integer, is.double (or just
is.numeric) as well as is.logical and is.raw, all the above can be tested
with is.na() first to add up how many Na they contain. If then it is of
type character, you can add any blank strings.

So the algorithm would initialize y to sum(is.na(vec)) and then if the
vec is character, add the sum of how many empty strings.

Alternately, the function should deal with what it wants to do if any
other type is encountered. You can internally converts many things to
integer or character and then operate on them. Or you can return a zero or
raise an alarm when given something else.

In this case, simply setting y to zero before using it would make it
defined and avoid the error, albeit report nothing found if it was a double
or Boolean vector even if it did contain NA.


-----Original Message-----
From: R-help <r-help-bounces at r-project.org> On Behalf Of David Carlson
via R-help
Sent: Tuesday, August 9, 2022 11:33 AM
To: Erin Hodgess <erinm.hodgess at gmail.com>
Cc: r-help at r-project.org
Subject: Re: [R] Odd behavior of a function within apply

Could you have columns that are not character or integer so that y is
never defined in the function?

count1a(1:5/3)
Error in count1a(1:5/3) : object 'y' not found

David Carlson


On Mon, Aug 8, 2022 at 1:35 PM Erin Hodgess <erinm.hodgess at gmail.com>
wrote:

OK.?? I'm back again.?? So my test1.??df is 236x390 If I put in the
following:?? lapply(test1.??df,count1a) Error in FUN(X[[i]],
.??.??.??) :?? object 'y' not found > lapply(test1.??df,count1a) Error
in FUN(X[[i]],
.??.??.??) :?? object 'y' not found > sapply(test1.??df,count1a)
ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This
message came from outside your organization.

ZjQcmQRYFpfptBannerEnd

OK.  I'm back again.

So my test1.df is 236x390

If I put in the following:
 lapply(test1.df,count1a)
Error in FUN(X[[i]], ...) : object 'y' not found

lapply(test1.df,count1a)

Error in FUN(X[[i]], ...) : object 'y' not found

sapply(test1.df,count1a)

Error in FUN(X[[i]], ...) : object 'y' not found

What am I doing wrong, please?
Thanks,
Erin


Erin Hodgess, PhD
mailto: erinm.hodgess at gmail.com


On Mon, Aug 8, 2022 at 1:41 PM Erin Hodgess <erinm.hodgess at gmail.com>

wrote:

Awesome, thanks so much!!

Erin Hodgess, PhD
mailto: erinm.hodgess at gmail.com


On Mon, Aug 8, 2022 at 1:38 PM John Fox <jfox at mcmaster.ca> wrote:

Dear Erin,

The problem is that the data frame gets coerced to a character
matrix, and the only column with "" entries is the 9th (the second
one you
supplied):

as.matrix(test1.df)
    X1_1_HZP1 X1_1_HBM1_mon X1_1_HBM1_yr
1  "48160"   "December"    "2014"
2  "48198"   "June"        "2018"
3  "80027"   "August"      "2016"
4  "48161"   ""            NA
5  NA        ""            NA
6  "48911"   "August"      "1985"
7  NA        "April"       "2019"
8  "48197"   "February"    "1993"
9  "48021"   ""            NA
10 "11355"   "December"    "1990"

(Here, test1.df only contains the three columns you provided.)

A solution is to use sapply:

 > sapply(test1.df, count1a)

     X1_1_HZP1 X1_1_HBM1_mon  X1_1_HBM1_yr
             2             3             3


I hope this helps,
  John


On 2022-08-08 1:22 p.m., Erin Hodgess wrote:

Hello!

I have the following data.frame
  dput(test1.df[1:10,8:10])
structure(list(X1_1_HZP1 = c(48160L, 48198L, 80027L, 48161L, NA,
48911L, NA, 48197L, 48021L, 11355L), X1_1_HBM1_mon =
c("December", "June", "August", "", "", "August", "April",
"February", "", "December"), X1_1_HBM1_yr = c(2014L, 2018L,
2016L, NA, NA, 1985L, 2019L, 1993L, NA, 1990L)), row.names =
c(NA, 10L), class = "data.frame")

And the following function:

dput(count1a)

function (x)
{
     if (typeof(x) == "integer")
         y <- sum(is.na(x))
     if (typeof(x) == "character")
         y <- sum(x == "")
     return(y)
}
When I use the apply function with count1a, I get the following:
  apply(test1.df[1:10,8:10],2,count1a)
     X1_1_HZP1 X1_1_HBM1_mon  X1_1_HBM1_yr
            NA             3            NA
However, when I do use columns 8 and 10, I get the correct response:
  apply(test1.df[1:10,c(8,10)],2,count1a)
    X1_1_HZP1 X1_1_HBM1_yr
            2            3

I am really baffled.  If I use count1a on a single column, it
works

fine.

Any suggestions much appreciated.
Thanks,
Sincerely,
Erin


Erin Hodgess, PhD
mailto: erinm.hodgess at gmail.com

      [[alternative HTML version deleted]]

https://urldefense.com/v3/__http://www.R-project.org/posting-guide.
html__;!!KwNVnqRv!CHx9JKnbOObpAt0LltEogLSxDUEl9qJDI6FgqMJBG_kdRHAy8
SJJdx6Uq0p4rpBa4E3DkmQ65UImH48MdYOqruE$

--
John Fox, Professor Emeritus
McMaster University
Hamilton, Ontario, Canada
web:
https://urldefense.com/v3/__https://socialsciences.mcmaster.ca/jfox
/__;!!KwNVnqRv!CHx9JKnbOObpAt0LltEogLSxDUEl9qJDI6FgqMJBG_kdRHAy8SJJ
dx6Uq0p4rpBa4E3DkmQ65UImH48MRU4wu3o$

      [[alternative HTML version deleted]]

        [[alternative HTML version deleted]]