R：在两个矩阵中合并公共行

Question

I have two matrices. I would like to combine them such that any element which is in Matrix1 but not in Matrix2 (scenario1) is added to the end of Matrix2. However, if an element is in both Matrix1 and Matrix2 (scenario2), then I would like to overwrite certain columns in that row of Matrix2 with the columns for the corresponding row of Matrix1.

I have taken a look at ddply and merge, which seem to satisfy scenario1, but I can't solve the problem regarding scenario2.

An example:
The original matrices: Matrix2

 Col1 Col2 Col3 Col4 ABC 100 200 900 DEF 300 400 1000 

Matrix1

 Col1 Col2 Col3 HIJ 500 600 ABC 700 800 KLM 1100 1200 

The new Matrix2:

 Col1 Col2 Col3 Col4 ABC 700 800 900 DEF 300 400 1000 HIJ 500 600 0 KLM 1100 1200 0 

Where the first row in the original Matrix2 has been replaced by the corresponding row from Matrix1 and the second rows from Matrix1 and Matrix2 have both been appended at the end. The extra column in Matrix2 (Col4) is unadjusted when combining the matrices. Also, the matrices have different dimensions.

Any help would be great!

Thanks

Mike

Answer 1

Perhaps a solution with rbind and duplicated might be of use:

m1m2 <- rbind(Matrix2, Matrix1)
m1m2[!duplicated(m1m2$Col1), ]
#   Col1 Col2 Col3
# 1  HIJ  500  600
# 2  ABC  700  800
# 4  DEF  300  400

Is the resulting order of "Col1" important?

---

Update

Based on your update, perhaps you can look for an option from the "reshape2" package combining melt , merge , duplicated , and dcast :

library(reshape2)
M1 <- melt(Matrix1, id.vars="Col1") ## Convert your data into a "long" format
M2 <- melt(Matrix2, id.vars="Col1")
M1M2 <- merge(M1, M2, all = TRUE)   ## Merge this long data
dcast(M1M2[!duplicated(M1M2[1:2], fromLast=TRUE), ], Col1 ~ variable, fill=0)
#   Col1 Col2 Col3 Col4
# 1  ABC  700  800  900
# 2  DEF  300  400 1000
# 3  HIJ  500  600    0
# 4  KLM 1100 1200    0

My guess is that you might need to add another variable in here to identify the source data.frame to ensure you are taking the correct data in the last step.