robustgp 2.0

Robust Gaussian process regression based on iterative trimming.

robustgp

Robust Gaussian Process Regression Based on Iterative Trimming

Zhao-Zhou Li, Lu Li, Zhengyi Shao, 2020, https://arxiv.org/abs/2011.11057

First application

• Modeling Unresolved Binaries of Open Clusters in the Color-Magnitude Diagram. I. Method and Application of NGC 3532 https://ui.adsabs.harvard.edu/abs/2020ApJ...901...49L/

Quick start

You can start with examples in this notebook.

Dependency

GPy

Install

Install the dependency first

pip install GPy

Assuming you want to put the code at certain directory, say ~/pythonlib

cd ~/pythonlib
wget https://raw.githubusercontent.com/syrte/robustgp/master/robustgp/robustgp.py

Then you can import ITGP as following,

import sys
sys.path.append('~/pythonlib')
from robustgp import ITGP

Or add this in your .bashrc once for all

export PYTHONPATH="$HOME/pythonlib:$PYTHONPATH"

I will write a setup.py in the future for easier installation.

Usage

from robustgp import ITGP

# train ITGP
res = ITGP(X, Y, alpha1=0.5, alpha2=0.975, nsh=2, ncc=2, nrw=1)
gp, consistency = res.gp, res.consistency

# make prediction
y_avg, y_var = gp.predict(x_new)
y_var *= consistency

See this notebook for a complete example.

Here gp is a GPy.core.GP object, whose usage is illustrated here.

License

The MIT License