Table 1 — tbl_one()

The bread-and-butter call: stratified baseline characteristics with auto-selected tests, SMDs, missing summaries, and an overall column.

Anatomy

ps.tbl_one(
    data,                   # pandas or polars DataFrame
    by="arm",               # stratification column (optional)
    variables=[...],        # which columns to include + order
    labels={...},           # display labels
    types={...},            # override automatic variable typing
    nonnormal=[...],        # use median (Q1, Q3) + rank tests
    tests={...},            # per-variable test overrides
    digits=2,
    pct_digits=1,
    missing="ifany",        # "ifany" | "always" | "never"
)

Variable typing

PySofra classifies each column as continuous, categorical, dichotomous, or ordinal:

• bool → dichotomous
• exactly {0, 1} ints → dichotomous
• few-level small ints → categorical
• string with 2 levels → dichotomous, more → categorical
• ordered pd.Categorical → ordinal
• numeric → continuous

Override with types={'smoker': 'categorical'}.

Test selection

Variable kind	Default test
Continuous, 2 groups	Welch's t-test
Continuous, 3+ groups	One-way ANOVA
Continuous + nonnormal=[...]	Wilcoxon / Kruskal–Wallis
Categorical, 2×2	Fisher's exact
Categorical, larger	Pearson χ²

Override per variable: tests={'age': 'wilcoxon', 'race': 'fisher'}.

Modifiers

Method	Effect
.add_p()	p-value column
.add_q(method='fdr_bh')	q-value column (multiplicity-adjusted)
.add_smd()	standardised mean difference column
.add_overall(label='Overall')	unstratified column
.bold_p(threshold=0.05)	bold rows below threshold
.theme('clinical')	apply a theme
.set_caption(text) / .add_footnote(text)	annotations

All modifiers return a new SofraTable (immutable).