An interactive Streamlit application for comparing the means of two independent samples using confidence intervals and hypothesis testing.
| Histogram | Results | Confidence Intervals |
|---|---|---|
 |
 |
 |
Designed for analysts, researchers and students, the toolkit automates common statistical calculations while providing clear visualisations and explanations of the results.
The repository also contains the original Jupyter notebook used during development for those interested in the underlying methodology.
🚀 Run online https://ab-testing-toolkit.streamlit.app
📓 Run notebook: notebooks/ab_testing.ipynb
💻 Run locally
Clone the repository
git clone https://github.com/steviecurran/ab-testing-toolkit.git
cd ab-testing-toolkit
Create a virtual environment
python3 -m venv .venv
source .venv/bin/activate
Install the dependencies
pip install -r requirements.txt
Launch Streamlit
streamlit run app/ab_testing_app.py
📊 Compare two independent sample means
📈 Automatic selection of z or t statistics
📐 Equal variance (pooled) or Welch's unequal variance test
🎯 One- or two-tailed hypothesis tests
📏 Adjustable confidence level
📂 Load data from: -repository datasets - local CSV/DAT files - URL - summary statistics
📉 Histogram comparison of sample distributions
📍 Publication-style confidence interval plots
📋 Plain-English interpretation of statistical results
.
├── app
│ └── ab_testing_app.py
├── src
│ └── statistics.py
├── data
├── notebooks
├── assets
└── README.md
###Example 1 – Magnesium supplement study
The repository contains a small example dataset (Mg_levels.dat) comparing magnesium levels before and after supplementation.
The histogram illustrates the distributions.
Using summary statistics only, the toolkit compares systolic blood pressure for men and women.
Although the difference in means is relatively small, the large sample size results in a statistically significant difference.
The application supports
- Independent two-sample t-test
- Welch's t-test
- z approximation for large samples
- Confidence intervals
- Hypothesis testing
- Effect estimation
The original notebook used to develop the toolkit is available in
notebooks/ab_testing.ipynb
It documents the development process and provides additional explanation of the underlying statistical methods.
- Effect size measures (Cohen's d)
- Power analysis
- Paired t-test
- Proportion testing
- Bootstrap confidence intervals
MIT Licence