R 循环或申请从数据框或列表中提取矩阵

Question

I am using trade data between countries and would like to create a country by country matrix per item traded. A question like this has been asked before here and I am using that persons example data, it hasn't been answered, my question is different as I would not necesarily need to do this for multiple years. DATA

usage <- data.frame(RC=c("DE", "IT", "USA","BRA", "ARG"),
                    Value=c(3,2,4,77,9,2,6,7,5,4),
                    Item = rep(c("Almonds", "Apples"), each=5))
import<- data.frame(RC=c("DE", "IT", "USA"),
                        PC = c("BRA", "ARG"),
                        Item = c("Almonds", "Apples"),
                        Value = c(1,5,3,2,8,3))

So I need to end up with ~450 identical matrixes for different products but all in the same matrix with all the countries included(so countries not trading a product should be included with value zero in that products matrix). Ultimately I would also like to fill the diagonal of these matrixes with the usage data from countries. I dont know if this is far too complex but i hope its possible. My code unfortunately does not work.....

Any help in the right direction would be much appreciated.

Edit: Solved by @user2474226. I filled the diagonals by adding the usage data as trade data before the matrixes are made.

Final code

usage$PC <- usage$RC #Duplicating the country column of usage
import2 <- rbind(import, usage) # adding the usage data to the trade data

import_YI <- split(import2, list(import2$Item))

trade_matrices <-lapply(import_YI, function(d) {
  x <- with(d, tapply(Value, list(RC, PC), sum)); 
  x[is.na(x)] <- 0; 
  x <- as.matrix(x);
  un1 <- unique(sort(c(colnames(x), rownames(x))));
  m2 <- matrix(0, NROW(un1), NROW(un1), dimnames = list(un1, un1));
  m2[rownames(x), colnames(x)] <- x;
  m2
  }
)

Many thanks to @user2474226 I learned a whole lot more and the solutions were very helpful. First time asking here, so didn't expect such a quick solution! Desired Output matrix per product

Answer 1

You could do something like this:

import_YI <- split(import, list(import$Item))
lapply(import_YI, function(d) {
                     x <- with(d, tapply(Value, list(RC, PC), sum)); 
                     x[is.na(x)] <- 0; 
                     x})
#output
$Almonds
    ARG BRA
DE    0   1
IT    0   8
USA   0   3

$Apples
    ARG BRA
DE    2   0
IT    5   0
USA   3   0

I'm converting each dataframe in the list from long form to wide form via the tapply . Where there's no traded product between the pair of countries, I'll get an NA, which I then replace with 0 .

Update : To create square matrices incorporating all the countries in rows and columns (using :

lapply(import_YI, function(d) {
    x <- with(d, tapply(Value, list(RC, PC), sum)); 
    x[is.na(x)] <- 0; 
    x <- as.matrix(x);
    un1 <- unique(sort(c(colnames(x), rownames(x))));
    m2 <- matrix(0, NROW(un1), NROW(un1), dimnames = list(un1, un1));
    m2[rownames(x), colnames(x)] <- x;
    m2}
    )
# output
$Almonds
    ARG BRA DE IT USA
ARG   0   0  0  0   0
BRA   0   0  0  0   0
DE    0   1  0  0   0
IT    0   8  0  0   0
USA   0   3  0  0   0

$Apples
    ARG BRA DE IT USA
ARG   0   0  0  0   0
BRA   0   0  0  0   0
DE    2   0  0  0   0
IT    5   0  0  0   0
USA   3   0  0  0   0