dabest 2025.3.27

Data Analysis and Visualization using Bootstrap-Coupled Estimation.

DABEST-Python

Recent Version Update

We are proud to announce DABEST Version Dadar (v2025.03.27) This new version of the DABEST Python library includes several new features and performance improvements. It’s a big one!

1. Python 3.13 Support: DABEST now supports Python 3.10—3.13.

2. Horizontal Plots: Users can now create horizontal layout plots, providing compact data visualization. This can be achieved by setting horizontal=True in the .plot() method. See the Horizontal Plots tutorial for more details.

3. Forest Plots: Forest plots provide a simple and intuitive way to visualize many delta-delta (or delta g), mini-meta, or regular delta effect sizes at once from multiple different dabest objects without presenting the raw data. See the Forest Plots tutorial for more details.

4. Gridkey: Users can now represent experimental labels in a ‘gridkey’ table. This can be accessed with the gridkey parameter in the .plot() method. See the gridkey section in the Plot Aesthetics tutorial for more details.

5. Other Visualization Improvements:

   • Comparing means and effect sizes: The estimation plots now include three types of customizable visual features to enhance contextualization and comparison of means and effect sizes:

     • Bars for the mean of the observed values (raw_bars): Colored rectangles that extend from the zero line to the mean of each group’s raw data. These bars visually highlight the central tendency of the raw data.

     • Bars for effect size/s (contrast_bars): Similar to raw bars, these highlight the effect-size difference between two groups (typically test and control) in the contrast axis. They provide a visual representation of the differences between groups.

     • Summary bands (reference_band): An optional band or ribbon that can be added to emphasize a specific effect size’s confidence interval that is used as a reference range across the entire contrast axis. Unlike raw and contrast bars, these span horizontally (or vertically if horizontal=True) and are not displayed by default.

       Raw and contrast bars are shown by default. Users can customize these bars and add summary bands as needed. For detailed customization instructions, please refer to the Plot Aesthetics tutorial.

   • Tighter spacing in delta-delta and mini-meta plots: We have adjusted the spacing of delta-delta and mini-meta plots to reduce whitespace. The new format brings the overall effect size closer to the two-groups effect sizes. In addition, delta-delta plots now have a gap in the zero line to separate the delta-delta from the ∆ effect sizes.

   • Delta-delta effect sizes for proportion plots: In addition to continuous data, delta-delta plots now support binary data (proportions). This means that 2-way designs for binary outcomes can be analyzed with DABEST.

   • Proportion plots sample sizes: The sample size of each binary option for each group can now be displayed. These can be toggled on/off via the prop_sample_counts parameter.

   • Effect size lines for paired plots: Along with lines connecting paired observed values, the paired plots now also display lines linking the effect sizes within a group in the contrast axes. These lines can be toggled on/off via the contrast_paired_lines parameter.

   • Baseline error curves: To represent the baseline/control group in the contrast axes, it is now possible to plot the baseline dot and the baseline error curve. The dot is shown by default, while the curve can be toggled on/off via the show_baseline_ec parameter. This dot helps make it clear where the baseline comes from i.e. the control minus itself. The baseline error curve can be used to show that the baseline itself is an estimate inferred from the observed values of the control data.

   • Delta text: Effect-size deltas (e.g. mean differences) are now displayed as numerals next to their respective effect size. This can be toggled on/off via the delta_text parameter.

   • Empty circle color palette: A new swarmplot color palette modification is available for unpaired plots via the empty_circle parameter in the .plot() method. This option modifies the two-group swarmplots to have empty circles for the control group and filled circles for the experimental group.

6. Miscellaneous Improvements & Adjustments

   • Numba for speed improvements: We have added Numba to speed up the various calculations in DABEST. Precalculations will be performed during import, which will help speed up the subsequent loading and plotting of data.

   • Terminology/naming updates: During the refactoring of the code, we have made several updates to the documentation and terminology to improve clarity and consistency. For example:

     • Plot arguments have been adjusted to bring more clarity and consistency in naming. Arguments relating to the rawdata plot axis will now be typically referred to with raw while arguments relating to the contrast axis will be referred to with contrast. For example, raw_label replaces swarm_label and bar_label. The various kwargs relating to each different type of plot (e.g., swarmplot_kwargs) remain unchanged.

     • The method to utilise the Delta g effect size is now via the .hedges_g.plot() method rather than creating a whole new Delta_g object as before. The functionality remains the same, it plots hedges_g effect sizes and then the Delta g effect size alongside these (if a delta-delta experiment was loaded correctly).

   • Updated tutorial pages: We have updated the tutorial pages to reflect the new features and changes. The tutorial pages are now more comprehensive and (hopefully!) more intuitive!

   • Results dataframe for delta-delta and mini-meta plots: A results dataframe can now be extracted for both the delta-delta and mini-meta effect size data (similar to the results dataframe for the regular effect sizes). These can be found via the .results attribute of the .delta_delta or .mini_meta object.

Contents

• About
• Installation
• Usage
• How to cite
• Bugs
• Contributing
• Acknowledgements
• Testing
• DABEST in other languages

About

DABEST is a package for Data Analysis using Bootstrap-Coupled ESTimation.

Estimation statistics are a simple framework that avoids the pitfalls of significance testing. It employs familiar statistical concepts such as means, mean differences, and error bars. More importantly, it focuses on the effect size of one’s experiment or intervention, rather than succumbing to a false dichotomy engendered by P values.

An estimation plot comprises two key features.

1. It presents all data points as a swarm plot, ordering each point to display the underlying distribution.

2. It illustrates the effect size as a bootstrap 95% confidence interval on a separate but aligned axis.

DABEST powers estimationstats.com, allowing everyone access to high-quality estimation plots.

Installation

This package is tested on Python 3.11 and onwards. It is highly recommended to download the Anaconda distribution of Python in order to obtain the dependencies easily.

You can install this package via pip.

To install, at the command line run

pip install dabest

You can also clone this repo locally.

Then, navigate to the cloned repo in the command line and run

pip install .

Usage

import pandas as pd
import dabest

# Load the iris dataset. This step requires internet access.
iris = pd.read_csv("https://github.com/mwaskom/seaborn-data/raw/master/iris.csv")

# Load the above data into `dabest`.
iris_dabest = dabest.load(data=iris, x="species", y="petal_width",
                          idx=("setosa", "versicolor", "virginica"))

# Produce a Cumming estimation plot.
iris_dabest.mean_diff.plot();

Please refer to the official tutorial for more useful code snippets.

How to cite

Moving beyond P values: Everyday data analysis with estimation plots

Joses Ho, Tayfun Tumkaya, Sameer Aryal, Hyungwon Choi, Adam Claridge-Chang

Nature Methods 2019, 1548-7105. 10.1038/s41592-019-0470-3

Paywalled publisher site; Free-to-view PDF

Bugs

Please report any bugs on the issue page.

Contributing

All contributions are welcome; please read the Guidelines for contributing first.

We also have a Code of Conduct to foster an inclusive and productive space.

A wish list for new features

If you have any specific comments and ideas for new features that you would like to share with us, please read the Guidelines for contributing, create a new issue using Feature request template or create a new post in our Google Group.

Acknowledgements

We would like to thank alpha testers from the Claridge-Chang lab: Sangyu Xu, Xianyuan Zhang, Farhan Mohammad, Jurga Mituzaitė, and Stanislav Ott.

Testing

To test DABEST, you need to install pytest and nbdev.

• Run pytest in the root directory of the source distribution. This runs the test suite in the folder dabest/tests/mpl_image_tests.
• Run nbdev_test in the root directory of the source distribution. This runs the value assertion tests in the folder dabest/tests

The test suite ensures that the bootstrapping functions and the plotting functions perform as expected.

For detailed information, please refer to the test folder

DABEST in other languages

DABEST is also available in R (dabestr) and Matlab (DABEST-Matlab).