Exploring data

A first look at a dataset, all in one theme: distributions, categories, correlations, cumulative distributions and a missing-data map. Every figure below is produced by running the code shown.

import depictr as dp

wb = dp.wellbeing_survey()
ld = dp.lexical_decision()

Cumulative distribution by group

An empirical cumulative distribution, with the legend tucked into the bottom-right the curves leave empty once they saturate.

p = dp.ecdf_plot(ld, "RT", group="condition", legend_inside=True)
print(show(p))

A categorical comparison

One categorical column as a proportion within the levels of another, here education across the regions.

p = dp.explore_categorical(wb, "education", group="region")
print(show(p))

Two variables, any types

explore_bivariate chooses the plot from the column types, here a numeric response across a category.

p = dp.explore_bivariate(ld, "condition", "RT")
print(show(p))

For two numeric columns it draws a scatter with a fitted trend, and a group splits the trend line.

cy = dp.crop_yield()
p = dp.scatter_trend(cy, "fertiliser", "yield", group="treatment")
print(show(p))

Correlation heatmap

Every pairwise Pearson correlation among the numeric columns, on one diverging scale running from -1 to 1.

p = dp.correlation_heatmap(wb)
print(show(p))

Scatter-plot matrix

The pairwise relationships the heatmap condenses, drawn in full.

p = dp.explore_pairs(cy, cols=["rainfall", "fertiliser", "soil_ph", "yield"])
print(show(p, width=8, height=7))

Raincloud

Density, box and raw points together.

p = dp.raincloud_plot(ld, "RT", group="condition")
print(show(p))

Missing-data map

Columns are ordered most- to least-missing, so the legend sits inside the top-right over the most-complete columns.

p = dp.missingness_map(wb, legend_inside=True)
print(show(p))

Group means over the raw data

Each group's mean with a confidence interval, drawn over its jittered observations.

p = dp.group_comparison_plot(ld, "RT", "condition")
print(show(p))

Ridgeline

Stacked densities, one ridge per region, ordered by median.

p = dp.ridgeline_plot(wb, "life_satisfaction", "region")
print(show(p))

Dumbbell

Two age groups across the regions, each pair joined by a segment whose length is the gap between them.

import numpy as np

wb_age = wb.assign(age_group=np.where(wb["age"] < 50, "under 50", "50 or over"))
p = dp.dumbbell_plot(wb_age, "region", "life_satisfaction", "age_group",
                     legend_inside=True)
print(show(p))

Outliers

A single variable's box plot with points beyond 1.5 times the interquartile range flagged in the accent colour.

p = dp.outlier_plot(cy, "yield")
print(show(p))

A descriptive summary table

A 'Table 1' of means, counts and missingness by group, returned as a data frame ready for any formatter. Variables with missing values gain their own row.

tab = dp.summary_table(
    wb,
    vars=["life_satisfaction", "income", "stress", "education"],
    group="region")
print(table(tab))

variable	statistic	Overall	East	North	South	West
N		300	76	70	74	80
life_satisfaction	Mean (SD)	4.8 (0.9)	5.1 (0.8)	5.2 (0.9)	4.5 (1.0)	4.6 (0.8)
	Missing, n (%)	8 (3%)	3 (4%)	1 (1%)	4 (5%)	0 (0%)
income	Mean (SD)	59844.7 (16793.5)	62480.1 (17120.8)	67736.6 (16403.5)	53668.0 (13730.9)	55987.8 (16334.2)
	Missing, n (%)	11 (4%)	2 (3%)	2 (3%)	3 (4%)	4 (5%)
stress	Mean (SD)	5.1 (2.2)	4.4 (1.8)	4.0 (2.2)	5.9 (2.1)	5.8 (1.9)
education	postgraduate	46 (15%)	9 (12%)	11 (16%)	12 (16%)	14 (18%)
	secondary	123 (41%)	30 (39%)	25 (36%)	31 (42%)	37 (46%)
	undergraduate	131 (44%)	37 (49%)	34 (49%)	31 (42%)	29 (36%)