CODEWITHSUNDEEP
stringrsubset
user3322865
user3322865

Reputation: 63

Subset a dataframe by a variable number of specific columns R

this one has been bugging me for a couple of days now, and I havent had any luck on stack exchange yet. Essentially, I have two tables, one table defines what columns (by column number) to select from the second table. My initial plan was to string together the columns and pass that into a subselect statement, however when I define the string as as.character it's not happy, i.e.:

# Data Sets, Variable_Selection: Table of Columns to Select from Variable_Table

VARIABLE_SELECTION <- data.frame(Set.1 = c(3,1,1,1,1), Set.2 = c(0,3,2,2,2), Set.3 = c(0,0,3,4,3),
                                 Set.4 = c(0,0,0,5,4), Set.5 = c(0,0,0,0,5))

VARIABLE_TABLE <- data.frame(Var.1 = runif(100,0,10), Var.2 = runif(100,-100,100), Var.3 = runif(100,0,1),
                             Var.4 = runif(100,-1000,1000), Var.5 = runif(100,-1,1), Var.6 = runif(100,-10,10))

# Sting rows into character string of columns to select

VARIABLE_STRING <- apply(VARIABLE_SELECTION,1,paste,sep = ",",collapse = " ")
VARIABLE_STRING <- gsub(" ",",",VARIABLE_STRING)
VARIABLE_STRING <- data.frame(VAR_STRING = gsub(",0","",VARIABLE_STRING))

# Will actually be part of lapply function but, one line selection for demonstration:

VARIABLE_SINGLE_SET <- as.character(VARIABLE_STRING[4,])

# Subset table for selected columns

VARIABLE_TABLE_SUB_SELECT <- VARIABLE_TABLE[,c(VARIABLE_SINGLE_SET)]

#  Error Returned:
#  Error in `[.data.frame`(VARIABLE_TABLE, , c(VARIABLE_SINGLE_SET)) : 
#  undefined columns selected

I know the text formatting is the problem but I can't find a workaround, any suggestions?

Upvotes: 1

Views: 818

Answers (2)

Henrik
Henrik

Reputation: 67818

Does this give the desired result?

lapply(VARIABLE_SELECTION, function(x) VARIABLE_TABLE[ , x[x != 0], drop = FALSE])

Produces a list where each element is a subset of 'VARIABLE_TABLE' given by 'VARIABLE_SELECTION' (using a 'VARIABLE_TABLE' with fewer rows).

# $Set.1
#       Var.3    Var.1  Var.1.1  Var.1.2  Var.1.3
# 1 0.09536403 5.593292 5.593292 5.593292 5.593292
# 2 0.09086404 6.339074 6.339074 6.339074 6.339074
# 
# $Set.2
#        Var.3    Var.2  Var.2.1  Var.2.2
# 1 0.09536403 65.81870 65.81870 65.81870
# 2 0.09086404 66.79157 66.79157 66.79157
# 
# $Set.3
#        Var.3     Var.4    Var.3.1
# 1 0.09536403 -674.6672 0.09536403
# 2 0.09086404 -576.7986 0.09086404
# 
# $Set.4
#        Var.5     Var.4
# 1  0.5155411 -674.6672
# 2 -0.9593219 -576.7986
# 
# $Set.5
#        Var.5
# 1  0.5155411
# 2 -0.9593219

Upvotes: 1

agstudy
agstudy

Reputation: 121608

You should avoid sub-setting by number of columns and process by variables names or at least keep your index as integer list( no need to coerce to a string)

First To stay in the same idea, this correct your code. Basciaclly I coerce your variable to vector:

VARIABLE_TABLE[,as.numeric(unlist(strsplit(
        VARIABLE_SINGLE_SET,',')))]

Upvotes: 1

Related Questions