Discrepancy Analysis: Sequenzo and TraMineR Mapping

This page maps Sequenzo discrepancy-analysis functions to their closest R counterparts. Most names below are from TraMineR core; dissindic() is from TraMineRextras.

What This Section Covers

This section focuses on discrepancy-based sequence analysis: overall discrepancy, pseudo-ANOVA on dissimilarities, weighted permutation tests, multifactor models, regression trees, and window-wise group comparisons along the time axis.

For TraMineR users, this is the diss* family plus the seqdiff and seqtree workflows described in Studer et al. (2011). Distance matrices are usually built first with seqdist() in TraMineR or get_distance_matrix() in Sequenzo.

Main Mapping Table

Sequenzo function	R counterpart	Notes
overall_discrepancy	TraMineR::dissvar()	Overall discrepancy from a distance matrix; squared=False uses v = 1 and squared=True uses v = 2.
single_factor_association	TraMineR::dissassoc()	Single-factor pseudo-ANOVA on distances; returns pseudo F, pseudo Fbf, pseudo R², Bartlett, and Levene summaries.
marginal_factor_association	repeated TraMineR::dissassoc()	One dissassoc() per covariate column; raw marginal association, not Type II partial effects.
multifactor_association	TraMineR::dissmfacw()	Multifactor model with Type II-style partial contributions after other covariates are included.
distance_multifactor_anova	TraMineR::dissmfacw()	Lower-level multifactor engine behind multifactor_association().
gower_matrix	internal Gower-centering step in TraMineR::dissmfacw()	Centered Gower matrix used for distance-based multifactor ANOVA.
individual_indicators	TraMineRextras::dissindic()	Individual marginality and gain indicators; not exported from TraMineR core.
merge_cluster_groups	TraMineR::dissmergegroups()	Greedy merging of cluster labels by partition-quality loss.
compare_groups_across_positions	TraMineR::seqdiff()	Window-wise local discrepancy scan along the time axis.
print_group_differences_across_positions	print() on a TraMineR seqdiff object	Text summary of a seqdiff object; TraMineR registers this as an S3 method.
plot_group_differences_across_positions	plot() on a TraMineR seqdiff object	Plot statistics or discrepancy profiles across anchor positions; TraMineR registers this as an S3 method.
distance_tree	TraMineR::disstree()	Distance-based regression tree on a distance matrix plus predictors.
sequence_tree	TraMineR::seqtree()	Sequence-facing wrapper that can compute the distance matrix before calling the tree fitter.
test_tree_split	internal split-significance logic behind TraMineR::disstree()	Permutation test for one candidate tree split.
get_leaf_membership	TraMineR::disstreeleaf()	Leaf ID or readable path label for each sequence.
get_classification_rules	TraMineR::disstree.get.rules()	Human-readable rules for terminal nodes.
assign_to_leaves	TraMineR::disstree.assign()	Assign new profiles to fitted leaves using tree rules.
print_tree	print() on TraMineR disstree / seqtree objects	Text display of a fitted tree; TraMineR registers these as S3 methods.
plot_tree	TraMineR::disstreedisplay() / TraMineR::seqtreedisplay()	GraphViz-based tree display for fitted distance-tree or sequence-tree objects.
export_tree_to_dot	TraMineR::disstree2dot() / TraMineR::seqtree2dot()	Export a tree to Graphviz DOT format.
permutation_test	TraMineR internal permutation wrapper	Generic permutation wrapper aligned with TraMineR's internal permutation logic.
association_permutation_test	TraMineR internal dissassocweighted.* helpers	Five-statistic dissassoc permutation engine used by single_factor_association().

Parameter Mapping for Core Entry Points

Sequenzo argument	TraMineR argument	Applies to
distance_matrix	diss	overall_discrepancy, single_factor_association, distance_tree, multifactor helpers
group	group / fac	single_factor_association
weights	weights	All weighted discrepancy functions
R	R	Association tests, trees, and multifactor routines that support permutation
weight_permutation	weight.permutation	single_factor_association, distance_tree, sequence_tree
squared	squared	All routines that accept a nonsquared (v = 1) versus squared (v = 2) dissimilarity transform
predictors	formula right-hand predictors	distance_tree, sequence_tree
min_size	min.size	distance_tree, sequence_tree
max_depth	maxdepth	distance_tree, sequence_tree
pval	pval	distance_tree, sequence_tree
first_split	first.split	distance_tree, sequence_tree
seqdata	left-hand formula term in seqtree	sequence_tree, compare_groups_across_positions
distance_method / distance_params	seqdist.args	sequence_tree
seqdist_args	seqdist.args	compare_groups_across_positions
cmprange	cmprange	compare_groups_across_positions
with_missing	with.missing	compare_groups_across_positions
weighted	weighted	sequence_tree, compare_groups_across_positions
factors	formula predictors in dissmfacw	multifactor_association, marginal_factor_association
gower	gower	multifactor_association, individual_indicators

TraMineR tree functions also accept max.depth and first as formal argument names. This guide uses maxdepth and first.split to match the Studer et al. (2011) examples and the other pages in this section.

Important Workflow Differences

• Import surface. Import discrepancy functions from sequenzo.discrepancy_analysis only. Internal folders such as stats/, trees/, positionwise/, and internal/ are for maintainers.
• Distance input. TraMineR often passes a dist object or matrix through a formula. Sequenzo usually takes an explicit distance_matrix, except in sequence_tree() and compare_groups_across_positions().
• Weighted permutation default. Sequenzo resolves weight_permutation=None the same way TraMineR does for dissassoc(), disstree(), and seqtree(): "none" without weights and "replicate" when weights are supplied. Unweighted calls also resolve to "none" inside TraMineR dissassocweighted() and DTNdisstree(). For survey or calibration weights, pass weight_permutation="diss" explicitly, as recommended by Studer et al. (2011). The seqdiff workflow calls dissassoc(..., weight.permutation = "diss") for local window scans, and compare_groups_across_positions() follows that path.
• Marginal versus multifactor. marginal_factor_association() is not a substitute for dissmfacw. Use multifactor_association() when you need conditional Type II-style contributions after other covariates are included.
• Position-wise scan. compare_groups_across_positions() rebuilds local distance matrices inside a moving window. It is not a cell-by-cell state composition comparison unless the window length is effectively one.
• Representative and factor plots. TraMineR also provides disscenter, dissrep, dissrf, dissdomassoc, disstreedisplay, and seqtreedisplay. Sequenzo does not expose one-to-one public wrappers for those helpers in this module.

Related TraMineR Functions Without a Direct Sequenzo Wrapper Here

TraMineR function	Typical role
TraMineR::disscenter()	Gravity center / medoid search from a distance matrix
TraMineR::dissrep()	Representative sequence selection from distances
TraMineR::dissrf()	Relative-frequency groups from distances
TraMineR::dissdomassoc()	Association between domain-specific distance matrices
TraMineR::disstreedisplay() / TraMineR::seqtreedisplay()	Rich tree displays beyond the basic plot_tree() helper

Authors

Code and documentation: Yuqi Liang

Reference

Studer, M., Ritschard, G., Gabadinho, A., & Müller, N. S. (2011). Discrepancy analysis of state sequences. Sociological Methods & Research, 40(3), 471–510.