Installation¶
Basic install¶
pip install splita
This installs splita with its only required dependencies: numpy and scipy.
Optional extras¶
splita ships optional extras for additional functionality. Install them as needed:
pip install splita[ml]
Adds scikit-learn for ML-powered variance reduction (CUPAC, CausalForest, HTEEstimator).
pip install splita[viz]
Adds matplotlib for built-in plotting of results, confidence intervals, and power curves.
pip install splita[widget]
Adds ipywidgets for interactive experiment dashboards in Jupyter notebooks.
pip install splita[api]
Adds FastAPI and uvicorn for serving splita as a REST API via splita.serve().
pip install splita[ml,viz,widget,api]
Development install¶
For contributing to splita:
git clone https://github.com/Naareman/splita.git
cd splita
pip install -e ".[dev,ml]"
This installs splita in editable mode with dev tools: pytest, ruff, mypy, and pytest-cov.
Requirements¶
- Python 3.10 or later
- numpy >= 1.24
- scipy >= 1.10
Verify installation¶
import splita
print(splita.__version__)
from splita import Experiment
result = Experiment([0, 1, 0], [1, 1, 1]).run()
print(result.method) # 'ztest'