Reshaping and summing dataframe values in R-CodePudding

I have a dataframe (full) like this:

obtained by merging a list of numeric ID codes and a dataframe (reproducible example):

library(plyr)
library(dplyr)

#Create example list: ID codes

l1 <- c(10, 100, 1500)
l2 <- c(45, 100)
l3 <- c(100, 1500, 3000, 4000)

l <- list(l1, l2, l3)

#Convert list into dataframe
ldf <- ldply(l, rbind)

#Create example dataframe
i <- data.frame(index = c(1, 3, 5))

#Merge the two dataframes
full <- merge(i, ldf, by = 'row.names', all = TRUE) %>% select(-Row.names)

I would like to reshape the dataframe as follows:

in order to get the sum of the index values for each ID code.

Any idea??

CodePudding user response：

in Base R, we can aggregate the values once we stack them:

aggregate(index~values, cbind(full['index'],stack(full,-1)), sum)

  values index
1     10     1
2     45     3
3    100     9
4   1500     6
5   3000     5
6   4000     5

Using tidyverse:

library(tidyverse)

full %>% 
   pivot_longer(-index, values_drop_na = TRUE) %>%
   group_by(value) %>%
   summarise(sum_index = sum(index))

# A tibble: 6 x 2
  value sum_index
  <dbl>     <dbl>
1    10         1
2    45         3
3   100         9
4  1500         6
5  3000         5
6  4000         5

CodePudding user response：

Here is a data.table approach:

library(data.table)

melt(setDT(full),id="index",na.rm=T)[, .(Sum.index = sum(index)), by=.(Cell.ID=value)]

Output:

   Cell.ID Sum.index
1:      10         1
2:      45         3
3:     100         9
4:    1500         6
5:    3000         5
6:    4000         5