Association measures between domains in multidimensional sequence analysis

Computes various measures of association between dimensions of multidimensional sequence data.

assoc.domains(dlist, names, djsa)

Arguments

dlist

A list of dissimilarity matrices or dist objects (see dist), with one element per dimension of the multidimensional sequence data

names

A character vector of the names of the dimensions of the multidimensional sequence data

djsa

A dissimilarity matrix or a dist object (see dist), corresponding to the distances between the multimdimensional sequences

References

Piccarreta R. (2017). Joint Sequence Analysis: Association and Clustering, Sociological Methods and Research, Vol. 46(2), 252-287.

Author

Nicolas Robette

Examples

# \donttest{
library(TraMineR)
data(biofam)

## Building one channel per type of event (left, children or married)
bf <- as.matrix(biofam[, 10:25])
children <-  bf==4 | bf==5 | bf==6
married <- bf == 2 | bf== 3 | bf==6
left <- bf==1 | bf==3 | bf==5 | bf==6

## Building sequence objects
child.seq <- seqdef(children)
#>  [>] 2 distinct states appear in the data: 
#>      1 = FALSE
#>      2 = TRUE
#>  [>] state coding:
#>        [alphabet]  [label]  [long label] 
#>      1  FALSE       FALSE    FALSE
#>      2  TRUE        TRUE     TRUE
#>  [>] 2000 sequences in the data set
#>  [>] min/max sequence length: 16/16
marr.seq <- seqdef(married)
#>  [>] 2 distinct states appear in the data: 
#>      1 = FALSE
#>      2 = TRUE
#>  [>] state coding:
#>        [alphabet]  [label]  [long label] 
#>      1  FALSE       FALSE    FALSE
#>      2  TRUE        TRUE     TRUE
#>  [>] 2000 sequences in the data set
#>  [>] min/max sequence length: 16/16
left.seq <- seqdef(left)
#>  [>] 2 distinct states appear in the data: 
#>      1 = FALSE
#>      2 = TRUE
#>  [>] state coding:
#>        [alphabet]  [label]  [long label] 
#>      1  FALSE       FALSE    FALSE
#>      2  TRUE        TRUE     TRUE
#>  [>] 2000 sequences in the data set
#>  [>] min/max sequence length: 16/16

## Using Hamming distance
mcdist <- seqdistmc(channels=list(child.seq, marr.seq, left.seq),
   method="HAM")
#>  [!!] 3  domains with  2000  sequences
#>  [>] building MD sequences of combined states...
#>  OK
#>  [>] computing substitution cost matrix for domain 1
#>  [>] computing substitution cost matrix for domain 2
#>  [>] computing substitution cost matrix for domain 3
#>  [>] computing MD substitution and indel costs with additive trick...
#>  OK
#>  [>] computing distances using additive trick ...
#>  [>] 2000 sequences with 7 distinct states
#>  [>] checking 'sm' (size and triangle inequality)
#>  [>] 537 distinct  sequences 
#>  [>] min/max sequence lengths: 16/16
#>  [>] computing distances using the HAM metric
#>  [>] elapsed time: 0.499 secs
child.dist <- seqdist(child.seq, method="HAM")
#>  [>] 2000 sequences with 2 distinct states
#>  [>] creating a 'sm' with a single substitution cost of 1
#>  [>] creating 2x2 substitution-cost matrix using 1 as constant value
#>  [>] 38 distinct  sequences 
#>  [>] min/max sequence lengths: 16/16
#>  [>] computing distances using the HAM metric
#>  [>] elapsed time: 0.321 secs
marr.dist <- seqdist(marr.seq, method="HAM")
#>  [>] 2000 sequences with 2 distinct states
#>  [>] creating a 'sm' with a single substitution cost of 1
#>  [>] creating 2x2 substitution-cost matrix using 1 as constant value
#>  [>] 58 distinct  sequences 
#>  [>] min/max sequence lengths: 16/16
#>  [>] computing distances using the HAM metric
#>  [>] elapsed time: 0.459 secs
left.dist <- seqdist(left.seq, method="HAM")
#>  [>] 2000 sequences with 2 distinct states
#>  [>] creating a 'sm' with a single substitution cost of 1
#>  [>] creating 2x2 substitution-cost matrix using 1 as constant value
#>  [>] 64 distinct  sequences 
#>  [>] min/max sequence lengths: 16/16
#>  [>] computing distances using the HAM metric
#>  [>] elapsed time: 0.237 secs

## Association between domains
asso <- assoc.domains(list(child.dist,marr.dist,left.dist), c('child','marr','left'), mcdist)
asso
#> $correlations
#> $correlations$pearson
#>       child  marr  left   jsa
#> child 1.000 0.339 0.067 0.673
#> marr  0.339 1.000 0.117 0.679
#> left  0.067 0.117 1.000 0.653
#> jsa   0.673 0.679 0.653 1.000
#> 
#> $correlations$spearman
#>       child  marr  left   jsa
#> child 1.000 0.283 0.052 0.605
#> marr  0.283 1.000 0.119 0.658
#> left  0.052 0.119 1.000 0.653
#> jsa   0.605 0.658 0.653 1.000
#> 
#> 
#> $`mean squared correlations`
#> $`mean squared correlations`$pearson
#> [1] 0.447
#> 
#> $`mean squared correlations`$spearman
#> [1] 0.408
#> 
#> 
#> $`Cronbach's alpha`
#> $`Cronbach's alpha`$`(child,marr,left)`
#> [1] 0.388
#> 
#> $`Cronbach's alpha`$`(child,marr)`
#> [1] 0.506
#> 
#> $`Cronbach's alpha`$`(child,left)`
#> [1] 0.126
#> 
#> $`Cronbach's alpha`$`(marr,left)`
#> [1] 0.209
#> 
#> 
# }