Getting started

Toolkit for developing, optimising and evaluating Likelihood Ratio (LR) systems. This allows benchmarking of LR systems on different datasets, investigating impact of different sampling schemes or techniques, and doing case-based validation and computation of case LRs.

LIR was first released in 2020 and redesigned from scratch in 2025, replacing the previous repository.

Installation

LIR is compatible with Python 3.12 and later. The easiest way to install LIR is to use pip:

pip install lir

For more detailed instructions of the CLI please refer to the project README.md.

Usage

This repository offers both a Python API and a command-line interface.

Command-line interface

To evaluate an LR system using the command-line interface, define your experiments in a YAML file and run lir:

lir <yaml file>

The examples folder may be a good starting point for setting up an experiment.

The elements of the experiment configuration YAML are looked up in the registry. The following lists all available elements in the registry.

lir --list-registry

Contributing

If you want to contribute to the LiR project, please follow the CONTRIBUTING.md guidelines, which include the instructions to set up LiR for local development.