CODEWITHSUNDEEP
rif-statementrow
dede
dede

Reputation: 1171

ifelse with multiple outputs

Is it possible to obtain multiple outputs with ifelse? I would like to create a variable ExpectedVariable with either 0s or 1s. If VariableA has a value of 0, then I need ExpectedVariable to be 0 for the current row and for the previous row. If VariableA has a value of 1, then I need ExpectedVariable to be 1.

I was thinking of something like this (which as it is doesn't work):

 mydata$ExpectedVariable <- ifelse(mydata$VariableA == 0, (mydata$ExpectedVariable & lag(mydata$ExpectedVariable) <- 0), 1)

Example:

         VariableA    ExpectedVariable
row1     1            1
row2     1            1
row3     1            1
row4     1            1
row5     1            1
row6     1            0
row7     0            0
row8     1            1
row9     1            0
row10    0            0
row11    1            1
row12    1            1
row13    1            0
row14    0            0

Upvotes: 0

Views: 1787

Answers (2)

Julien Navarre
Julien Navarre

Reputation: 7830

Should do the trick :

df$expectedVariable[df$VariableA == 1] <- 1
df$expectedVariable[df$VariableA == 0] <- 0
df$expectedVariable[which(df$VariableA == 0) - 1] <- 0

Or with the 2 last lines in one :

df$expectedVariable[c(which(df$VariableA == 0), which(df$VariableA == 0) - 1)] <- 0

Upvotes: 0

duckmayr
duckmayr

Reputation: 16920

Another way you can think about your problem is that ExpectedVariable will equal zero if either of two conditions are met (current value of VariableA is zero, or the lead of VariableA is zero). In other words, instead of thinking of multiple outputs, think of multiple conditions, which ifelse() can easily do. Then we can accomplish this with any sort of shifting function.

Using data.table::shift():

expected_df <- read.table(header = TRUE, text = "
         VariableA    ExpectedVariable
row1     1            1
row2     1            1
row3     1            1
row4     1            1
row5     1            1
row6     1            0
row7     0            0
row8     1            1
row9     1            0
row10    0            0
row11    1            1
row12    1            1
row13    1            0
row14    0            0")

rownames(expected_df) <- NULL

mydata <- data.frame(VariableA = expected_df$VariableA)
mydata$ExpectedVariable <- ifelse(mydata$VariableA == 0
                                  | data.table::shift(mydata$VariableA,
                                                      type = 'lead') == 0,
                                  0, 1)
all.equal(mydata, expected_df)

[1] TRUE

Using dplyr::lead()

mydata$ExpectedVariable <- ifelse(mydata$VariableA == 0
                                  | dplyr::lead(mydata$VariableA) == 0,
                                  0, 1)

all.equal(mydata, expected_df)

[1] TRUE

Upvotes: 1

Related Questions