Finding certain rows and determining if they are negative or positive-CodePudding

I am trying to do something quite odd. I would like to somehow determine if the row of a separate data.frame is mostly negative or positive based on row numbers in a column of a different data.frame... I have included my example data frames (data.frame1 and data.frame2) and the desired output

>data.frame1
     col_a col_b  col_c
 1   1     1       1
 2   1    -1      -1 
 3   1     1       1 
 4  -1    -1      -1 
 5   1     1      -1

>data.frame2
   col_a
1    2
2    3
3    5


Example output/result 
     col_a     col_b
1      2     negative
2      3     positive
3      5     positive

Another example output could be

Example 2 output/result 
     col_a  col_b
1      2    -1
2      3     1
3      5     1

CodePudding user response：

Subset first dataframe rows based on second dataframe, then check the sign, then get the sum for the rows, then get sign again:

cbind(data.frame2,
      result = sign(rowSums(sign(data.frame1[ data.frame2$col_a, ]))))
#   col_a result
# 1     2     -1
# 2     3      1
# 3     5      1

Note: if the first dataframe has only values -1 and 1, then we can drop the inner sign step:

sign(rowSums(data.frame1[ data.frame2$col_a, ]))

CodePudding user response：

This is my solution using rowMeans instead of RowSums.

#making the example
df1 <- data.frame("a" = c(1,1,1,-1,1),
                  "b" = c(1,-1,1,-1,1),
                  "c"=  c(1,-1,1,-1,-1))

df2 <- data.frame("col_a" = c(2,3,5))

# Check if all indices of rows in d2 are in df1
stopifnot(all(df2$col_a %in% c(1:nrow(df1))))

# no second sign to get a logical value
df2$col_b <- rowMeans(sign(df1[df2$col_a,])) >= 0

# make to factor
df2$col_b <- ifelse(df2$col_b,"positive","negative")