Visualization Tools

Sequenzo's plots fall into two families, and the sidebar groups them this way:

• Longitudinal views keep individual trajectories intact, with one visual element per sequence: index plot, most frequent sequences, relative frequency, single medoid, and longitudinal characteristics.
• Cross-sectional views summarize the sample at each time point: state distribution, modal state, mean time, transition matrix, and cross-sectional characteristics.

How to Read Sequence Plots explains how to interpret each plot type and choose between them. The Gallery shows every plot with code.

Common Parameters and How Plots Work

The rest of this page covers the mechanics shared by all plot functions. If you understand these, you can use any plot with confidence.

Where the plot appears

• Jupyter Notebook → the figure shows under the cell.
• Python script → a window opens (depends on your environment).
• If a function has show, setting show=False suppresses on-screen display; you can still save via save_as.

Saving figures

save_as="figure.png"   # or "figure.pdf", "figure.jpg"
dpi=200                # 300+ for publication, if your machine can handle it

• If you omit the extension, .png is used.
• The file goes to your current working directory unless you provide an absolute path.

Figure size

figsize=(width, height) in inches.

• Many groups or long time spans → larger figures (e.g., (14, 8)).
• Quick previews → smaller figures render faster.

Colors and legends

• Colors come from your SequenceData object and stay consistent across plots.

• Legends also come from SequenceData (same labels, same colors).

• Want to tweak a color? Edit the palette before plotting, for example:

  # Example: change the color for the "Education" state
  seqdata.color_map_by_label["Education"] = "#1f77b4"

Time axis and units

• Time labels are taken from SequenceData (e.g., years/months/waves).
• Make sure axis labels match your unit (e.g., “Year”, “Month”, “Wave”).
• Some functions use percentages (0–100%), others use proportions (0–1). Axis labels tell you which it is.

Grouping into panels

• id_group_df must map each “Entity ID” to a group.

• categories is the column in id_group_df that names the group variable.

• Layout:

  • layout="column" stacks groups vertically.
  • layout="grid" arranges groups in a grid; optionally set nrows/ncols.

• Order:

  • group_order fixes the panel order.
  • sort_groups controls automatic ordering (“auto”, “numeric”, “alpha”, “none”).

Sorting sequences (where applicable)

• Some plots sort rows to reveal structure:

  • Lexicographic by time (default) or rules like "transition_count", "final_state", etc.

• Sorting changes only the display order, not your data.

Performance tips

• Big datasets (many sequences/time points) → use show=False and save to file; open the image later.
• Prefer aggregated plots (e.g., modal state, distribution) when the index plot is too dense.
• Use vector formats (.pdf, .svg) for print; use .png for slides and docs.

Quick presets (copy/paste)

# Publication-ready PNG
save_as="figure.png"; dpi=300; figsize=(14, 8)

# Fast preview in Notebook
save_as=None; dpi=150; figsize=(10, 6)

# Batch export without popping windows
show=False; save_as="figure.png"

Troubleshooting

• Blank or tiny plot:

  • Data may not match groups. Check that id_group_df["Entity ID"] aligns with seqdata.ids.

• Colors look wrong or inconsistent:

  • Ensure state labels in your data match SequenceData labels; adjust color_map_by_label before plotting.

• No legend or cut-off legend:

  • Some functions place the legend outside the axes. Increase figure width/height or save to PDF.

• “Nothing saved”:

  • Check save_as path and extension. Without extension, .png is added.

Author: Yuqi Liang