[R] Dataframes: conditional calculations per row .

[David Winsemius]

One way. there may be better. The apply function will work with just  
one row (or one column) at a time.

 > DF
    Month Week Estpassage MedFL
1   July   27        665    34
2   July   28       2232    35
3   July   29       9241    35
4   July   30      28464    35
5    Aug   31      41049    35
6    Aug   32      82216    35
7    Aug   33     230411    35
8    Aug   34     358541    35
9   Sept   35     747839    35
10  Sept   36     459682    36
11  Sept   37     609567    36
12  Sept   38     979475    36
13  Sept   39     837189    36

Build a function, say we call it switch.cond:

  switch.cond <- function (x) {
                if (x["Week"] >= 33) return( fun1(x) )  else  
return( fun2(x) ) }

  # Build two more functions to handle the dispatched rows

  fun1 <-function(x){ cat("do function 1\n") }
# replace the cat-call with your first calculation

  fun2 <-function(x){ cat("do function 2\n") }
# and use various x["<colname>"]'s as arguments

 > apply(DF, 1, switch.cond)
do function 2
do function 2
do function 2
do function 2
do function 2
do function 2
do function 1
do function 1
do function 1
do function 1
do function 1
do function 1
do function 1
NULL

HTH:
David Winsemius



On Feb 9, 2009, at 5:11 PM, Jesús Guillermo Andrade wrote:

> Dear Sirs: I've been working with several variables in a dataframe
> that serve as part of a calculation that I need to perform in a
> different way depending on its value. Let me explain:
>
> The main dataframe is called llmcc
>
> llmcc : 'data.frame':	283 obs. of  11 variables:
>  $ Area     : num  308.8 105.6 51.4 51.4 52.9 ...
>  $ mFondo   : num  30.1 10 10.2 10.2 40.4 ...
>  $ mFachada : num  22.95 6.7 4.72 4.72 4.72 ...
>  $ Marca    : Factor w/ 132 levels "AA_Movilnet",..: 11 32 82 82 32
> 32 32 32 32 32 ...
>  $ Clase    : int  8 4 1 1 1 1 1 1 12 1 ...
>  $ Categoria: int  2 6 6 6 1 1 1 1 1 1 ...
>  $ Phi      : num  0.128 0.147 0.217 0.217 0.887 ...
>  $ Rf       : num  0.119 0.102 0.147 0.147 0.143 ...
>  $ OldA     : num  0.737 0.258 0.375 0.375 0.385 ...
>  $ OldCondo : num  4436 1555 2260 2260 2318 ...
>  $ NewA_Jon : num  1.069 0.368 0.256 0.256 0.264 ...
>
> I perform an initial operation using the original variables plus one
> numeric (Abase) that is external and has the same number of rows than
> the dataframe:
>
> alitemp  <- ((Abase/llmcc$Clase)*PClase)+(((1/llmcc
> $Categoria)*Abase)*PCategoria)+((Abase*llmcc$Phi)*PPhi)+((Abase*llmcc
> $Rf)*PRf)
>
> So, after I obtain the results of this calculation, I append the
> series by creating an additional column within the original dataframe:
> l
> lmcc$Alitmp  <- alitemp
>
> Problem is: I need to calculate a new column using a formula that has
> different structure depending on the values of llmcc$Clase, thus: for
> any given row of llmcc where llmcc$Clase is >= 10 i would have to
> perform some operations with other values in the same row that are, by
> definition, different than the ones I would need in case of lmcc$Clase
> is < 10.
> I've managed to break down the original dataframe by using subsets,
> and then performing the calculations, but then it is complicated to
> put the results in the same order of the original dataframe.
> I understand the workings of the control structures available in R but
> after reading the docs and help files, I can´t figure how to perform a
> conditional calculation row by row that checks first the values of a
> given column and then applies the corresponding operation to another
> column, so it outputs a series in the same exact order as the  
> dataframe.
>
> Any light that you might share with me over this will be highly
> appreciated.
>
> Thanks in advance.
>
>
> Guillermo.
>
>
>
> Nunca le preguntes a un peluquero si necesitas un corte de pelo. Ley
> de Murray.
> ------------------------------
> Jesús Guillermo Andrade (Abg.)
> Gerente de Litigios y Corporativo. EDM. AC. API.
> Andrade & Moreno S.C. (http://amlegal.wordpress.com/)
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.