plot_subsequence_group_contrasts()

plot_subsequence_group_contrasts() visualizes discriminating subsequences across groups.

It is Sequenzo's closest counterpart to TraMineR plot.subseqelistchisq.

Function Usage

plot_subsequence_group_contrasts(
    group_contrast_results,
    y_limit_mode="uniform",
    rows=None,
    cols=None,
    resid_levels=(0.05, 0.01),
    color_palette=None,
    plot_type="freq",
    legend_title=None,
    show_legend=True,
    legend_text_scale=1.0,
    figsize=(13, 7),
    fontsize=11,
    x_label=None,
    y_label=None,
    save_as=None,
    dpi=200,
    show=False
)

TraMineR Parameter Mapping

• group_contrast_results -> TraMineR x
• y_limit_mode -> TraMineR ylim
• plot_type -> TraMineR ptype
• show_legend -> TraMineR with.legend
• legend_text_scale -> TraMineR cex.legend
• x_label / y_label -> TraMineR xlab / ylab

Entry Parameters

Parameter	Required	Type	Description
group_contrast_results	✓	SubsequenceList	Group-comparison subsequence results, usually from compare_groups().
plot_type	✗	str	"freq" for frequency view, "resid" for Pearson residual view.
x_label, y_label	✗	str	Axis label overrides.
save_as	✗	str	Save path; .png is auto-appended if missing.
dpi	✗	int	Save resolution (default 200).
show	✗	bool	If True, calls plt.show() inside the function.

What It Does

• Produces group-wise panels for discriminating subsequences.
• Encodes over- or under-representation in each group using Pearson-residual-based color bins.
• Supports both frequency and residual-centric interpretation.

Example (Step by Step)

from sequenzo.event_sequences import plot_subsequence_group_contrasts

# Step 1: Frequency-style view
plot_subsequence_group_contrasts(
    discr[:10],
    plot_type="freq",
    x_label="Frequency",
    y_label="Subsequence",
    save_as="outputs/subsequence_group_contrasts_freq",
    dpi=300,
    show=True
)

# Step 2: Residual-style view
plot_subsequence_group_contrasts(
    discr[:10],
    plot_type="resid",
    x_label="Pearson residual",
    y_label="Subsequence",
    save_as="outputs/subsequence_group_contrasts_resid",
    dpi=300,
    show=True
)

R Counterpart

• Closest R function: plot.subseqelistchisq
• Mapping note: Residual-based color encoding follows the same visual logic as Figures 3 and 4 in Ritschard et al. (2013).

Authors

Code: Yuqi Liang

Documentation: Yuqi Liang

Reference

Ritschard, G., Bürgin, R., & Studer, M. (2013). Exploratory Mining of Life Event Histories. In J. J. McArdle & G. Ritschard (Eds.), Contemporary Issues in Exploratory Data Mining in the Behavioral Sciences (pp. 221–253). Routledge.