Installation

Requirements

GPClarity requires Python 3.9 or later. The core library has minimal dependencies, but full functionality requires GPy.

Basic Installation

For analysis and lightweight usage:

pip install gpclarity

This installs the core interpretability tools without heavy scientific computing dependencies.

Development Installation

To install from source with development tools:

git clone https://github.com/AngadKumar16/gpclarity.git
cd gpclarity
pip install -e ".[dev]"

Dependencies

Core dependencies:

• numpy >= 1.20

• scipy >= 1.7

• GPy >= 1.10

Optional dependencies (install separately as needed):

• matplotlib >= 3.4 — required for all plot methods

• pandas >= 1.3 — required for HyperparameterTracker.to_dataframe()

• joblib — enables parallel LOO computation in DataInfluenceMap

• tqdm — shows progress bars during LOO computation

• scikit-learn — used for nearest-neighbour extrapolation detection

Verifying Installation

import gpclarity
print(gpclarity.__version__)
print(f"Full features available: {gpclarity.AVAILABLE['full']}")