gt_plt_conf_int

gt_plt_conf_int(
    gt,
    column,
    ci_columns=None,
    ci=0.95,
    width=100,
    dot_color='red',
    border_color='red',
    line_color='royalblue',
    text_color='black',
    text_size='default',
)

Create confidence interval plots in GT cells.

The gt_plt_conf_int() function takes an existing GT object and adds horizontal confidence interval plots to a specified column. Each cell displays a horizontal bar representing the confidence interval, with a dot indicating the mean value. Optionally, the lower and upper confidence interval bounds can be provided directly, or the function can compute them.

If ci_columns is not provided, the function assumes each cell in column contains a list of values and computes the confidence interval using a t-distribution.

Parameters

gt : GT: A GT object to modify.
column : SelectExpr: The column that contains the mean of the sample. This can either be a single number per row, if you have calculated the values ahead of time, or a list of values if you want to calculate the confidence intervals.
ci_columns : SelectExpr | None = None: Optional columns representing the left/right confidence intervals of your sample. If None, the confidence interval will be computed from the data in column using a t-distribution.
ci : float = 0.95: The confidence level to use when computing the interval (if ci_columns is None).
width : float | int = 100: The width of the confidence interval plot in pixels.
dot_color : str = 'red': The color of the mean dot.
border_color : str = 'red': The color of the border around the mean dot.
line_color : str = 'royalblue': The color of the confidence interval bar.
text_color : str = 'black': The color of the confidence interval labels.
text_size : Literal['small', 'default', 'large', 'largest', 'none'] = 'default': The size of the text for the confidence interval labels. Options include "none" for no text.

Returns

: GT: A GT object with confidence interval plots added to the specified column.

Examples

import pandas as pd
from great_tables import GT
import gt_extras as gte

df = pd.DataFrame({
    'group': ['A', 'B', 'C'],
    'mean': [5.2, 7.8, 3.4],
    'ci_lower': [3.1, 6.1, 1.8],
    'ci_upper': [7.3, 9.7, 5.0],
    'ci': [5.2, 7.8, 3.4],
})

gt = GT(df)
gt.pipe(
    gte.gt_plt_conf_int,
    column='ci',
    ci_columns=['ci_lower', 'ci_upper'],
    width=120,
)

group	mean	ci_lower	ci_upper	ci
A	5.2	3.1	7.3	3.1 7.3
B	7.8	6.1	9.7	6.1 9.7
C	3.4	1.8	5.0	1.8 5

Alternatively we can pass in a column of lists, and the function will compute the CI’s for us.

import numpy as np
np.random.seed(37)

n_per_group = 50
groups = ["A", "B", "C"]
means = [20, 22, 25]
sds = [10, 16, 10]

# Create the data
data = []
for i, (grp, mean, sd) in enumerate(zip(groups, means, sds)):
    values = np.random.normal(mean, sd, n_per_group)
    data.extend([{"grp": grp, "values": val} for val in values])

df_raw = pd.DataFrame(data)
df_summary = (
    df_raw
    .groupby("grp")
    .agg({"values": ["count", "mean", "std", list]})
    .round(3)
)
df_summary.columns = ["n", "avg", "sd", "ci"]

gt = GT(df_summary)
gt.pipe(
    gte.gt_plt_conf_int,
    column="ci",
)

n	avg	sd	ci
50	21.391	9.99	19 24
50	21.719	16.927	16.91 26.53
50	25.528	10.522	23 29

Note

All confidence intervals are scaled to a common range for visual alignment.