CODEWITHSUNDEEP
rmathcorrelation
Евгений Греков
Евгений Греков

Reputation: 71

R Cross-correlation via multiple columns

I have a data table like this:

> head(my_data)
     V1    V2    V3    V4    V5
1 36045 49933 41622 29491 34393
2 36874 44752 44158 40561 36045
3 45008 51964 58015 32733 29491
4 44830 72017 60434 40347 40561
5 48553 65470 49933 38842 32733
6 52028 64955 44752 41622 40347

I have already learned how to find correlation via multiple columns:

> head(cor(my_data)[,])
          V1        V2        V3        V4        V5
V1 1.0000000 0.4621777 0.7985130 0.9490929 0.9045297
V2 0.4621777 1.0000000 0.8041824 0.4201712 0.1583757
V3 0.7985130 0.8041824 1.0000000 0.7466672 0.5889458
V4 0.9490929 0.4201712 0.7466672 1.0000000 0.8672321
V5 0.9045297 0.1583757 0.5889458 0.8672321 1.0000000

I've tried a lot, but could not reach my goal to find maximum absolute value for cross-correlation with ccf funtion among every pair. Thanks a lot in advance for all your answers!

Upvotes: 2

Views: 716

Answers (1)

Sathish
Sathish

Reputation: 12703

mat
#         V1    V2    V3    V4    V5
# [1,] 36045 49933 41622 29491 34393
# [2,] 36874 44752 44158 40561 36045
# [3,] 45008 51964 58015 32733 29491
# [4,] 44830 72017 60434 40347 40561
# [5,] 48553 65470 49933 38842 32733
# [6,] 52028 64955 44752 41622 40347

class(mat)
# [1] "matrix"

combins <- combn(colnames(mat), 2)

a1 <- apply(combins, 2,
            FUN = function(x){ccf(mat[, x[1]], mat[, x[2]])})

abs_max_ccf <- unlist(lapply(a1, function(x) abs(max(x$acf))))

names(abs_max_ccf) <- apply(combins, 2, function(x) paste0(x[1], x[2], collapse = ''))

abs_max_ccf
#      V1V2      V1V3      V1V4      V1V5      V2V3      V2V4      V2V5      V3V4      V3V5      V4V5 
# 0.7460529 0.4450512 0.5167570 0.4672099 0.8028452 0.4944933 0.5220862 0.4076768 0.2884272 0.8494897

Verify Results: Extract two columns from mat: V1 and V2 and perform absolute maximum of ccf.

abs(max(ccf(mat[, "V1"], mat[, "V2"])$acf))
# [1] 0.7460529

Upvotes: 1

Related Questions