define table of columns based on condition using R - r

Question

I have data which is stored in a dataframe as
CST M QS
501 1204315 1
501 1204324 1
501 6041541 2
501 7508226 1
501 7509677 1
501 7514622 1
503 7511601 2
507 6961200 2
507 7514100 2
507 7522029 1
509 7512374 1
509 7516799 2
511 419110 0.5
511 6000832 5
511 6960800 3.33
511 7010000 2
511 7508229 2.5
511 7508307 2
511 7515126 2
Now, i would like to change this data based on CST different M has to be stored in different columns. These columns should be dynamic such that CST can have only 1 M or it can be 10 or 20 or infinite that many columns has to be generated with M1, M2, M3 and so on. QS has to be displayed with sum of the values based on sum(M)
Sample output is as below
cst M1 M2 M3 M4 M5 M6 M7 Total
501 1204315 1204324 6041541 7508226 7509677 7514622 7
503 7511601 2
507 6961200 7514100 7522029 5
509 7512374 7516799 3
511 419110 6000832 6960800 7010000 7508229 7508307 7515126 17.33
I have used transpose() where it just converts rows to columns.But this will not resolve to the expected output which i want.
I have tried using data.table function as dt[,sum(QS),by="CST"] but am not able to display M along with it in different columns.
Thanks in advance.

Answer 1

We can use data.table. We convert the 'data.frame' to 'data.table' (setDT(df1)), grouped by 'CST', we get the sequence of rows (1:.N) and sum of "QS", assign (:=) the output to create two columns ("N", "Total"). Then, we use dcast to convert from long to wide format.
library(data.table)
setDT(df1)[, c("N", "Total") := list(paste0("M", 1:.N), sum(QS)), CST]
dcast(df1, CST+Total~N, value.var='M')
# CST Total M1 M2 M3 M4 M5 M6 M7
#1: 501 7.00 1204315 1204324 6041541 7508226 7509677 7514622 NA
#2: 503 2.00 7511601 NA NA NA NA NA NA
#3: 507 5.00 6961200 7514100 7522029 NA NA NA NA
#4: 509 3.00 7512374 7516799 NA NA NA NA NA
#5: 511 17.33 419110 6000832 6960800 7010000 7508229 7508307 7515126
Or we use the same methodology with dplyr/tidyr
library(dplyr)
library(tidyr)
df1 %>%
group_by(CST) %>%
mutate(Total=sum(QS), N=row_number()) %>%
select(-QS) %>%
spread(N, M)
Update
If we need the columns in the order, we can convert the "N" to factor with levels specified
setDT(df2)[, c("N", "Total") := list(paste0("M", 1:.N), sum(QS)), CST]
df2[, N:= factor(N, levels=unique(N))]
dcast(df2, CST+Total~N, value.var="M")
data
df2 <- structure(list(CST = c(501L, 501L, 501L, 501L, 501L,
501L, 501L,
501L, 501L, 501L, 501L, 501L, 503L, 507L, 507L, 507L, 509L, 509L,
511L, 511L, 511L, 511L, 511L, 511L, 511L), M = c(1204315L, 1204324L,
6041541L, 7508226L, 7509677L, 7434399L, 7843392L, 7834393L, 8343999L,
3439242L, 3434323L, 7514622L, 7511601L, 6961200L, 7514100L, 7522029L,
7512374L, 7516799L, 419110L, 6000832L, 6960800L, 7010000L, 7508229L,
7508307L, 7515126L), QS = c(1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1,
1, 2, 2, 2, 1, 1, 2, 0.5, 5, 3.33, 2, 2.5, 2, 2)),
.Names = c("CST",
"M", "QS"), class = "data.frame", row.names = c(NA, -25L))