Smoothing sequence data

Smoothing of sequence data, using for each sequence the medoid of the sequences in its neighborhood. The results can be used to get a smoothed index plot.

seqsmooth(seqdata, diss, k=20, r=NULL)

Arguments

seqdata

a sequence object (see seqdef function).

diss

a dissimilarity matrix, giving the pairwise distances between sequences.

k

size of the neighborhood. Default is 20.

r

radius of the neighborhood. If NULL (default), the radius is not used for smoothing.

Value

A list with the following elements:

seqdata

a sequence object (see seqdef function)

R2

pseudo-R2 measure of the goodness of fit of the smoothing

S2

stress measure of the goodness of fit of the smoothing

References

Piccarreta R. (2012). Graphical and Smoothing Techniques for Sequence Analysis, Sociological Methods and Research, Vol. 41(2), 362-380.

Author

Nicolas Robette

Examples

data(trajact)
seqact <- seqdef(trajact)
#>  [>] 6 distinct states appear in the data: 
#>      1 = 1
#>      2 = 2
#>      3 = 3
#>      4 = 4
#>      5 = 5
#>      6 = 6
#>  [>] state coding:
#>        [alphabet]  [label]  [long label] 
#>      1  1           1        1
#>      2  2           2        2
#>      3  3           3        3
#>      4  4           4        4
#>      5  5           5        5
#>      6  6           6        6
#>  [>] 500 sequences in the data set
#>  [>] min/max sequence length: 37/37
dissim <- seqdist(seqact, method="LCS")
#>  [>] 500 sequences with 6 distinct states
#>  [>] creating a 'sm' with a substitution cost of 2
#>  [>] creating 6x6 substitution-cost matrix using 2 as constant value
#>  [>] 377 distinct  sequences 
#>  [>] min/max sequence lengths: 37/37
#>  [>] computing distances using the LCS metric
#>  [>] elapsed time: 0.584 secs
mds <- cmdscale(dissim, k=1)
smoothed <- seqsmooth(seqact, dissim, k=30)$seqdata
seqIplot(smoothed, sortv=mds, xtlab=14:50, with.legend=FALSE, yaxis=FALSE, ylab=NA)