EvoCov extension to learn the kernel function.
Project description
GPlib extension to learn the kernel function.
Setup evocov
Create and activate virtualenv (for python2) or venv (for python3)
# for python3
python3 -m venv .env
# or for python2
python2 -m virtualenv .env
source .env/bin/activate
Upgrade pip
python -m pip install --upgrade pip
Install EvoCov package
python -m pip install evocov
Use EvoCov
Import EvoCov and GPlib to use it in your python script.
import gplib
import evocov
Configure the fitting method.
lml = gplib.me.LML()
bic = gplib.me.BIC()
fitting_method = evocov.fit.EvoCov(
obj_fun=bic.fold_measure,
max_fun_call=25000,
nested_fit_method=gplib.fit.MultiStart(
obj_fun=lml.fold_measure,
max_fun_call=250,
nested_fit_method=gplib.fit.LocalSearch(
obj_fun=lml.fold_measure,
method="Powell",
max_fun_call=100
)
)
)
Initialize the GP with None covariance function.
gp = gplib.GP(
mean_function=gplib.mea.Fixed(),
covariance_function=fitting_method.get_random_kernel()
)
Generate some random data.
import numpy as np
data = {
'X': np.arange(3, 8, 1.)[:, None],
'Y': np.random.uniform(0, 2, 5)[:, None]
}
Fit the kernel and the hyperparameters to the training set.
validation = gplib.dm.Full()
log = fitting_method.fit(gp, validation.get_folds(
data
))
Finally plot the posterior GP.
posterior_gp = gp.get_posterior(data)
gplib.plot.gp_1d(posterior_gp, data, n_samples=10)
There are more examples in examples/ directory. Check them out!
Develop EvoCov
Download the repository using git
git clone https://gitlab.com/ibaidev/evocov.git
cd evocov
git config user.email 'MAIL'
git config user.name 'NAME'
git config credential.helper 'cache --timeout=300'
git config push.default simple
Update API documentation
source ./.env/bin/activate
pip install Sphinx
cd docs/
sphinx-apidoc -f -o ./ ../evocov
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
evocov-0.0.8.tar.gz
(27.9 kB
view hashes)
Built Distribution
evocov-0.0.8-py2.py3-none-any.whl
(39.6 kB
view hashes)
Close
Hashes for evocov-0.0.8-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 4c2516af81073f988c0f747d0ddd67eff7a99e9b84c52cd1c6f10965bb95d940 |
|
MD5 | f9dc7107cb969968540f0fbaae8bda2b |
|
BLAKE2b-256 | 410fb524b90850f941b8bc3809c2e5318c9b3e538de08200006285c57e434ed9 |