Longitudinal-Toolkit CLI
Long-Toolkit (short for Longitudinal-Toolkit) is a hub of toolkits for longitudinal data in Python, from data preparation and preprocessing to modelling and explainability. It is designed for social, medical and clinical researchers who work with repeated-measure data and need a stepping-stone path from raw cohort files to downstream modelling.
The main toolkit is the benchmarking toolkit (benchmark_standard_ml, benchmark_longitudinal_ml and bench_standard_and_longitudinal_ml). It ranks standard and longitudinal machine learning under one shared protocol to establish whether machine learning, and/or longitudinal machine learning, is worth it on a given longitudinal dataset compared with a statistical baseline. The explainability stage then explains a saved model, e.g., the best estimator of a benchmark, with SHAP values.
The initiative delivers two interconnected components. First, the long-toolkit Python package is the engine behind every tool, fully usable from scripts or notebooks through its application programming interface (API). Second, this repository ships the long-toolkit command, a fully interactive command-line interface (CLI), written in Go, for running and orchestrating the toolkit from start to finish without code.
Each stage (data preparation, data preprocessing, machine learning and explainability) opens straight onto its tools, which help researchers iterate quickly on their datasets. The toolset is built to grow through community contributions.
Caution
This repository is the CLI component of the project. For the full end-to-end toolkit experience, workflows and primary documentation, head to long-toolkit and its documentation.
brew tap Long-Toolkit/homebrew-tap
brew install long-toolkit
scoop bucket add long-toolkit https://github.com/Long-Toolkit/scoop-bucket
scoop install long-toolkit
git clone https://github.com/Long-Toolkit/CLI.git
cd CLI
make build
Homebrew and Scoop put long-toolkit on PATH automatically.
When building from source, run the installer target for the shell in use:
# bash
make install-bash
# fish
make install-fishThen run it from any directory:
long-toolkitThe CLI drives the Python toolkit through a bridge and requires Python 3.10 only. Run it from a Python 3.10 project directory where the Python package is installed, for example after uv add long-toolkit (with uv, the state-of-the-art Python package manager).
We designed the CLI so that its UX stays smooth at every moment. The command palette is available from any screen and launches any tool instantly, without walking through each menu level by hand.
In practice, launch long-toolkit, open the palette with : or Ctrl+P, type a phrase like benchmark longitudinal ml, and press Enter to run it. Whilst typing, Tab auto-completes, Ctrl+H opens history, Ctrl+L clears input and Esc closes the palette.
Jump to the main repository and its documentation for workflows and end-user guidance:
Please do not share participant-level or restricted data in issues or pull requests.
Security policy and contact details:
Special thanks to @charm.land for their amazing TUI framework!