Skip to content

Welcome to Curve Fit!

Background

CurveFit is a primarily a Python package for fitting curves using nonlinear mixed effects. It can be used to do only that if desired. However, due to its former usage for the IHME COVID-19 project, it has modules specifically for evaluating model performance out beyond the range of time observed in the data. Likewise, it has modules for creating uncertainty intervals based on out of sample performance.

In our methods documentation we discuss the statistical methods for CurveFit. In our code documentation, we explain the core model code and also the extensions that allow for evaluating model performance and generating uncertainty intervals.

IHME COVID-19 Project

**This repository and method is no longer being used for the IHME COVID-19 project. IHME uses a SEIIR based approach, and the code can be found here:

For any IHME COVID-19 related inquiries, please contact covid19@healthdata.org.

To see the IHME projections visualization, click here. For FAQs, click here.

Getting Started

To clone the repository and get started, you can either do

git clone https://github.com/ihmeuw-msca/CurveFit.git
cd CurveFit
make install

A Note on cppad_py

One of the dependencies for this package is cppad_py, a python interface for algorithmic differentiation. If you experience issues installing or importing cppad_py after doing make install, please see this page to clone and debug the build for cppad_py with setup.py.

Maintainers