GSTools covariance model as variogram_model input #125

MuellerSeb · 2019-05-31T17:31:26Z

With this pull request an interface to GSTools is provided. The PyKrige routines UniversalKriging, UniversalKriging3D, OrdinaryKriging and OrdinaryKriging3D now allow an instance of the GSTools CovModel as input for variogram_model. Every information needed will be set internally.

An example was added to the README and to the example folder.

MuellerSeb · 2019-06-03T16:07:55Z

I went through the CI logs and I think the check-fails are not induced by the pull-request related changes.
Is there something broken in your CI setup or did I miss something?
Cheers,
Sebastian

rth

Thanks @MuellerSeb, this is very nice!

A few comments about the code,

no need to include examples/gstools_example.png, it will be generated while building the docs in https://pykrige.readthedocs.io/en/latest/examples/
CI is failing because gstools need to be installed in CI before examples are run.

It would be good to add optional tests to CI, something along the lines of,
pykrige/tests/test_core.py

@pytest.mark.paramterize(model, [OrdinaryKriging, UniversalKriging])
def test_gstools_variogram(model):
    gstools = pytest.importorskip("gstools")
  
    # create some data
    # create the gstools model
    # fit the krigging estimator

pykrige/ok.py

pykrige/ok3d.py

pykrige/uk.py

examples/gstools_covmodel.py

MuellerSeb · 2019-09-27T11:17:37Z

Thanks to @LSchueler I could resolve the diff issue. Now I can care about your suggestions. Sorry for the long delay, but I was a bit feared of the rebase.

MuellerSeb · 2019-11-13T21:43:26Z

Ping.

rth

Thanks! A few more comments. The first two (about factorizing code and linking in docstring) apply to all files where such changes are made.

rth · 2019-11-13T22:45:09Z

pykrige/ok.py

            which to rotate coordinate system in order to take into
            account anisotropy. Default is 0 (no rotation).
        """

+        # set up variogram model and parameters...
+        self.variogram_model = variogram_model


Can we factorize this code repeated in all estimators, maybe as,

def _setup_variogram_parameters(self, ..)

under pykrige.core. There is already an _initialize_variogram_model there. All that could probably be made methods of some base class, but that should probably be done in a separate PR..

I just tried that, but it turns out to be more difficult than thought. It should be done in a separate PR refactoring the kriging classes. The update_variogram_model method should be used here, but it has to be adopted, to work with different dimensions. I will leave it as it is now, if that is ok.

rth · 2019-11-13T22:46:08Z

pykrige/ok3d.py

        Specified which variogram model to use; may be one of the following:
        linear, power, gaussian, spherical, exponential, hole-effect.
        Default is linear variogram model. To utilize a custom variogram model,
        specify 'custom'; you must also provide variogram_parameters and
        variogram_function. Note that the hole-effect model is only technically
        correct for one-dimensional problems.
+        You can also use a GSTools
+        CovModel, see https://github.com/GeoStat-Framework/GSTools


Use RST link instead of raw url

`GSTools <https://github.com/GeoStat-Framework/GSTools>`_

rth · 2019-11-13T22:49:46Z

README.rst

@@ -31,6 +31,10 @@ point and all grid points. With the 'functional' drift capability, the user may
 of the spatial coordinates that define the drift(s). The package includes a module that contains functions
 that should be useful in working with ASCII grid files (`*.asc`).

+PyKrige also works smoothly together with the Covariance Model provided by
+`GSTools <https://github.com/GeoStat-Framework/GSTools>`_ v1.1 and higher.
+Have a look at the section: `GSTools covariance models`_.


I would remove this last sentence, as a link to the documentation of another package before the link to pykrige documentation is confusing. User who need this functionality would be able to find the instructions from other links in this PR.

rth · 2019-11-13T22:56:32Z

Also it looks like CI on master may also need fixing (the CI failures are likely unrelated to this PR). Would you be interested in opening a separate PR to look into it?

Sorry, this package is very little maintained: personally I'm no longer working on krigging related tasks, and we won't be able to merge this PR as long as CI is not green (at least for Python 3.5+).

MuellerSeb · 2019-11-14T10:49:13Z

@rth I can have a look at the CI and make a separate PR.

MuellerSeb · 2019-11-22T23:36:04Z

Also it looks like CI on master may also need fixing (the CI failures are likely unrelated to this PR). Would you be interested in opening a separate PR to look into it?

Done.

Sorry, this package is very little maintained: personally I'm no longer working on krigging related tasks, and we won't be able to merge this PR as long as CI is not green (at least for Python 3.5+).

Again, I could offer help with maintaining this package by integrating it to the GeoStat-Framework. Since it's an github organization for geostatics in python, it is easy to share maintainance of packages related to this topic.. at least that is what we are aiming at.
We can discuss about that if you want. Don't want to obtrude myself upon you.

MuellerSeb · 2019-11-25T13:28:20Z

should be ready now.

rth · 2019-12-02T21:41:55Z

I'll try to have a look later this week at the PR.

Again, I could offer help with maintaining this package by integrating it to the GeoStat-Framework. Since it's an github organization for geostatics in python, it is easy to share maintenance of packages related to this topic.. at least that is what we are aiming at.

Maybe try contacting @bsmurphy by email about it and cc me? At least to see if he could give you write access to this repo. Another solution could be to mark this repo as unmaintained with a link to a maintained version at that org. In any case, personally I won't have much availability to help with the maintenance of this package in the future, and it would be good if some solution could be found at least to merge the occasional PR from contributors (also a PhD is a good time to contribute to OSS :) )

MuellerSeb · 2019-12-09T15:36:16Z

ping.

MuellerSeb

Looks good to me.

MuellerSeb mentioned this pull request May 31, 2019

Interface with GSTools #124

Closed

rth reviewed Jun 9, 2019

View reviewed changes

pykrige/ok.py Outdated Show resolved Hide resolved

pykrige/ok3d.py Outdated Show resolved Hide resolved

pykrige/ok3d.py Outdated Show resolved Hide resolved

pykrige/uk.py Outdated Show resolved Hide resolved

examples/gstools_covmodel.py Outdated Show resolved Hide resolved

MuellerSeb added 2 commits September 27, 2019 12:02

added GSTools CovModel as possible variogram_model input; added example

eee42a2

README GSTools example

a407318

MuellerSeb added 4 commits September 27, 2019 13:51

raise ValueError for wrong dim; remove links in comments

be24d41

add test for gstools interface

bbaa7b2

update README gstools example

580b93b

update gstools example; remove png

a4f5ed7

rth reviewed Nov 13, 2019

View reviewed changes

MuellerSeb mentioned this pull request Nov 14, 2019

MAINT Fix CI setup #131

Merged

MuellerSeb mentioned this pull request Nov 22, 2019

Isotropic Semivariogram in pykrige #130

Closed

MuellerSeb added 8 commits November 25, 2019 11:03

Merge branch 'master' of github.com:bsmurphy/PyKrige

13b8ede

README: remove link to GSTools before PyKrige Documentation

0216442

GSTools example: apply CI hack for plotting

a7d9d96

GSTools: update link in DOC

b2483e9

CI: add gstools as dep

5ee3267

GSTools: fix example

f48cd4f

CI: gstools test only on travis and py36

b0bfec4

CI: skip additional installation if sklearn and gstools

5b61835

MuellerSeb mentioned this pull request Jan 12, 2020

20_pykrige_interface is broken GeoStat-Framework/GSTools#56

Closed

MuellerSeb mentioned this pull request Jan 28, 2020

GeoStat-Framework integration: PyKrige v2 #136

Open

10 tasks

MuellerSeb commented Jan 28, 2020

View reviewed changes

MuellerSeb merged commit 40a8140 into GeoStat-Framework:master Jan 28, 2020

MuellerSeb mentioned this pull request Apr 2, 2020

v1.5.0 #145

Merged

MuellerSeb mentioned this pull request Apr 12, 2021

Krige: Adapting to PyKrige GeoStat-Framework/GSTools#160

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

GSTools covariance model as variogram_model input #125

GSTools covariance model as variogram_model input #125

MuellerSeb commented May 31, 2019

MuellerSeb commented Jun 3, 2019

rth left a comment

MuellerSeb commented Sep 27, 2019

MuellerSeb commented Nov 13, 2019

rth left a comment

rth Nov 13, 2019

MuellerSeb Nov 25, 2019

rth Nov 13, 2019

rth Nov 13, 2019

rth commented Nov 13, 2019

MuellerSeb commented Nov 14, 2019

MuellerSeb commented Nov 22, 2019

MuellerSeb commented Nov 25, 2019

rth commented Dec 2, 2019 •

edited

Loading

MuellerSeb commented Dec 9, 2019

MuellerSeb left a comment

GSTools covariance model as variogram_model input #125

GSTools covariance model as variogram_model input #125

Conversation

MuellerSeb commented May 31, 2019

MuellerSeb commented Jun 3, 2019

rth left a comment

Choose a reason for hiding this comment

MuellerSeb commented Sep 27, 2019

MuellerSeb commented Nov 13, 2019

rth left a comment

Choose a reason for hiding this comment

rth Nov 13, 2019

Choose a reason for hiding this comment

MuellerSeb Nov 25, 2019

Choose a reason for hiding this comment

rth Nov 13, 2019

Choose a reason for hiding this comment

rth Nov 13, 2019

Choose a reason for hiding this comment

rth commented Nov 13, 2019

MuellerSeb commented Nov 14, 2019

MuellerSeb commented Nov 22, 2019

MuellerSeb commented Nov 25, 2019

rth commented Dec 2, 2019 • edited Loading

MuellerSeb commented Dec 9, 2019

MuellerSeb left a comment

Choose a reason for hiding this comment

rth commented Dec 2, 2019 •

edited

Loading