apply a function instead of using a loop

Question

I've been searching for a long time in the documentation and in forums, but I still have difficulties to understand how to use the apply function instead of a loop in R for more complex functions. (for function like apply(data, 1, sum) it ok though)

for example I have the following function

 AOV_GxT=function(Trial_group,trait,df){ 

        sub_table=df[which(df$Trial.group == Trial_group),]

        aov_GxT = anova(aov(sub_table[,trait] ~ Genotype + Treatment + Treatment/Rep.number + Genotype*Treatment, data=sub_table, na.action="na.omit"))

        pvalue = aov_GxT$"Pr(>F)"[2]
        return(c(Trial_group,trait,pvalue))  

    }

that I want to apply for each Trial_groups and each traits (in columns) from a dataframe df

so I usually do the following (that works perfectly) :

aov_table=data.frame(matrix(ncol=4))
colnames(aov_table)=c("Trial_group", "Trait", "pvalue G*T")
for(trait in colnames(dataset)[2:ncol(dataset)]){
  for(Trial_group in unique(dataset[,'Trial.group'])){
    aov_table<-cbind (aov_table,AOV_GxT(Trial_group,trait,dataset))  
  }

dataset is a dataframe containing the data and more columns containing factors for the aov in the function.

head(dataset)

     Trial.group     Trait1           Trait2           Trait3
          A        0.4709055        0.6123510        0.7098447
          B        0.4973123        0.6322532        0.7336145
          C        0.4955180        0.6243369        0.7336492
          D        0.4787380        0.6235426        0.7304343
          E        0.5137033        0.6418851        0.7364666
          F        0.4524246        0.5975655        0.7012825

I would like to limit the use of loops and learn how to use apply family functions, so I created lists and tried to use mapply :

trait_lst = list(colnames(df_vars_clean)[7:ncol(df_vars_clean)])
Tgrp_lst = list(unique(df_vars_clean[,'Trial.group']))

aov_table<-mapply(AOV_GxT(a,b,c),a=Tgrp_lst,b=trait_lst, c=dataset )

Then it gives me an error with the subset in the function, I think because I try to do a subset from a list:

object of type 'builtin' is not subsettable

I know that there may be some mistakes in my code, but I'm learning R all by myself and there are some notions that I don't understand well for the moment.

How could I use apply on my function instead of multiple for loops ?

Thanks.

maybe you need to write list(a=Tgrp_lst,b=trait_list,c=dataset) inside the mapply call. — Carl Witthoft
– Carl Witthoft, Commented Dec 10, 2013 at 15:56
@Charles It is better to try out ddply in your case. Given your data.frame (df) with colnames of "Trial_group","Trait","Genotype","Treatment","Rep.number" my.func <- funciton(sub_table) { anova(aov(Trait ~ Genotype + Treatment + Treatment/Rep.number + Genotype*Treatment, data=sub_table, na.action="na.omit")) }; res <- ddply(df,c("Trial_group","Trait"),function(x) df$pvalue=my.func(x)) — xb.
– xb., Commented Dec 10, 2013 at 16:23

jimmyb · Accepted Answer · 2013-12-10 16:49:53Z

2

As pointed out in the comments, ddply is good choice, however, the problem is easily solved by a lapply as well:

do.call(rbind, lapply(split(dataset, dataset$Trial.Group), function(tgDf) {
  do.call(rbind, lapply(c("Trait1", "Trait2", "Trait3"), function(trait) {
      ## you don't need the trial group, it is already subsetted.
      AOV_gtx(trait, tgDf)
  }))
}))

Using ddply you would remove the outer lapply/split code:

ddply(dataset, "Trial.Group", function(tgDf) {
   ## the code in here would be the same, because you are iterating over
   ## the response cols.
})

The key with all of these functions, and R in general, is don't preallocate datastructures to store the results -- it's functional so you are going to build up the results and then return them.

answered Dec 10, 2013 at 16:49

jimmyb

4,4072 gold badges25 silver badges26 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Charles Over a year ago

@xiaobei & jimmyb, thank you for your help, I didn't know the split and ddply functions. It works perfectly!

Collectives™ on Stack Overflow

apply a function instead of using a loop

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related