[R] How to avoid ifelse statement converting factor to character

[Craig P. Pyrame]

Dear Rolf,

Rolf Turner wrote:
>
> On 25/06/2009, at 12:27 PM, Craig P. Pyrame wrote:
>
>>
>> Dear Stavros,
>>
>> What you discuss below is somewhat scary to me as an R newbie.  Is this
>> just an incident, a bug perhaps, or rather the way things typically go
>> in R, as your "Welcome to R!" seems to suggest?  I have just started to
>> learn R, and my initial euphoria of the "I can do anything with it!"
>> sort is gradually turning into an "I can't get why it doesn't work" and
>> "I can't get how to make this work" depression.  I would be happy to
>> blame this on my incompetence and incapability, but would also like to
>> hear if it is not R itself that causes me to fail.
>
>     Do not get your knickers in a twist.  R works simply and 
> straightforwardly
>     in simple straightforward situations.  In less simple and less 
> straightforward
>     situations life gets more complicated.  Don't dive into such 
> situations
>     without making sure you understand them.  Check your results to 
> make sure
>     you have not overlooked a subtlety.
>

Yes, sure, but this sounds to me almost like what some others seem to 
have already suggested on this list - use R for tasks where you can 
simply apply any of the numerous excellent package functions (for 
example, to fit a model), but avoid using it for programming, because it 
inevitably leads to problems - is this what you are saying?

>     In respect of the current discussion of ifelse() --- the original 
> problem arose
>     because the values of ``yes'' and ``no'' were of different modes.  
> It is obvious
>     that in such instances a decision will have to be made about the 
> mode of the
>     result.  The appropriateness of the designers' decision may be 
> disputed, but
>     you have to admit that some decision had to be made.  Recognize 
> that and all
>     the mystery goes away.
>

I wasn't really updated on the original example, but yes, creating a 
programming language involves numerous difficult decisions that have to 
be made, often in a relatively ad hoc fashion.  But once you point me to 
the original example, I have looked into what Stavros wrote, and the 
following strikes me as one of those cases where I would feel that 
either me or R is wrong:

   > ifelse(TRUE,factor('x'),factor('x')) => 1      (integer)

The man page Stavros quotes states that the class attribute of the 
result is taken from 'test', which clearly is not the case:

   > class(TRUE) => logical
   > class(ifelse(TRUE,factor('x'),factor('x'))) => integer

And also, I find myself incapable of making sense of the "may" in "the 
mode of the result may depend on the value of 'test'" - may in what 
sense?  Will in 2.9 but not in 2.8?  Will if test is logical but not if 
it is raw?  I am not particularly into programming languages, and 
especially their semantics, so it may be my fault that I don't know what 
such "may" may mean, but I am not quite sure.  Could you help me?

>     If you don't understand what's going on, then just stick to using 
> ifelse() only
>     when ``yes'' and ``no'' have the same mode.

Yes, in my poor R programming I make efforts to avoid involving myself 
in tricky situations, but it is not always helpful enough, I assure you.

>
>     Using things like as.raw() or taking one of ``yes'' and ``no'' to 
> be a list is getting
>     into territory where you need to be quite sophisticated and quite 
> careful.  Unless
>     you are both, don't go there.  I consider myself to be both (in 
> respect of R at
>     least) and I *still* would be very reluctant to go there.
>

I am not sure, is this supposed to be an argument in favor of using R?


>     Bottom line:  R is easy to use at any level, but in order to use 
> it a ``high'' level
>     you need to understand the high level.  Don't attempt to run 
> before you can crawl.

I am a bit reluctant to tell you this, Rolf, as there is likely much my 
own fault involved, but I attempt to crawl, not to run, and this is 
where I get into pains with R.  Examples such as those of Stavros have 
no place in my code, and yet I find it surprisingly difficult to 
understand why large parts of my code don't work as expected - having 
read the man pages carefully enough, I'd think.

Best regards,
Craig