Skip to content

UQRegressors

UQRegressors is a Python package that provides machine learning regression models capable of generating prediction intervals for a user-specified confidence level. These models not only estimate the expected output but also quantify the uncertainty around each prediction. For instance, a model from UQRegressors trained to predict house prices could, given a new set of input features, return a 95% confidence interval—indicating that the true price is expected to lie between a predicted lower and upper bound with 95% certainty. Several models from Bayesian and Conformal Prediction literature are implemented and validated on a PyTorch backend, which can be easily applied to regression problems through a scikit-learn fit, predict interface.

Key Features

  • Highly customizable parameters for each model
  • Easy-to-use interface with built in dataset validation
  • GPU compatibility with PyTorch backend
  • Validated implementations with comparisons to published results
  • Easy saving and loading of created models
  • Wide variety of metrics available to assess quality of fit and prediction intervals

Use Cases

There are five main capabilities of UQRegessors:

  1. Dataset Loading and Validation
  2. Regression using models of various types created with UQ capability
  3. Hyperparameter Tuning using bayesian optimization (wrapper around Optuna)
  4. Metrics for evaluating goodness of fit and quality of uncertainty intervals
  5. Visualization of metrics, goodness of fit, and quality of uncertainty intervals

For a simple demonstration of how these features could be used for your problem, check the "Is UQRegressors right for me?" example. For a more holistic view of UQRegressors' cabilities, look at the "Getting Started" example.

Installation

To install all core components of UQRegressors, run: 

pip install UQRegressors

Installing PyTorch

UQRegressors requires PyTorch, which must be installed separately to match your system's configuration.

CPU-only:

pip install torch torchvision torchaudio

With CUDA support for GPU:

Choose the appropriate command based on your CUDA version:

  • CUDA 11.8: 
    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
  • CUDA 12.1: 
    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

For other versions or platforms, check the official PyTorch installation page

What Next?