pykrige

library1.7.3pypypi✓ verified 84d ago

PyKrige is a Kriging Toolkit for Python that supports 2D and 3D ordinary and universal kriging, as well as simple kriging, regression kriging, and conditional realizations. It provides a scikit-learn compatible interface and is developed by the GeoStat Framework project. Current version: 1.7.3. Release cadence: irregular, with several updates per year.

analytics maps

pip install pykrige

github ↗homepage ↗

INSTALL

IMPORT

SIG · PYKRIGE

pykrige

analyticspythonv1.7.3

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

OrdinaryKriging

✓ from pykrige.ok import OrdinaryKriging

✗ from pykrige import OrdinaryKriging

Subpackages ok, uk, etc. must be explicit; top-level import raises AttributeError.

UniversalKriging

✓ from pykrige.uk import UniversalKriging

✗ from pykrige import UniversalKriging

Same as above; use subpackage import.

Krige

✓ from pykrige.rk import RegressionKriging

RegressionKriging replaces older Krige; always use RegressionKriging.

Perform ordinary kriging on 2D synthetic data with a linear variogram model.

import numpy as np from pykrige.ok import OrdinaryKriging # Synthetic data np.random.seed(42) X = np.random.rand(20, 2) * 10 Z = np.sin(X[:, 0]) + np.cos(X[:, 1]) # Grid for prediction gridx = np.linspace(0, 10, 20) gridy = np.linspace(0, 10, 20) # Ordinary Kriging OK = OrdinaryKriging(X[:, 0], X[:, 1], Z, variogram_model='linear') z, ss = OK.execute('grid', gridx, gridy) print(z.shape, ss.shape)

Debug

Known issues

gotchaDo not use variogram_model='exponential' or 'gaussian' without checking data sparsity; they may produce singular matrix errors.

fix

Use 'linear' or 'spherical' as default; test other models on small datasets first.

affects: all

deprecatedThe old pykrige.kriging_tools module is deprecated. Use pykrige.tools instead.

fix

Replace 'from pykrige.kriging_tools import ...' with 'from pykrige.tools import ...'.

affects: >=1.6.0

gotchaWhen using the scikit-learn API, the fit() method requires additional keyword arguments beyond X, y (e.g., coordinates). Incorrect usage leads to unexpected errors.

fix

Always pass coordinate_data keyword: model.fit(X, y, coordinate_data=coords).

affects: all

Errors

Common errors & fixes

LinAlgError: Matrix is singular.

The kriging matrix (covariance matrix) is singular, often due to duplicate points or poorly chosen variogram model.

fix

Remove duplicate points using numpy.unique or adjust the variogram model (try 'linear').

AttributeError: module 'pykrige' has no attribute 'OrdinaryKriging'

Direct import from top-level pykrige, not from subpackage.

fix

Use 'from pykrige.ok import OrdinaryKriging'.

ValueError: The number of coordinates does not match the number of points.

Mismatch between coordinate array shape and data array shape in sklearn API.

fix

Ensure X is 2D (n_samples, n_features) and coordinate_data is 2D (n_samples, n_dims).

Upgrade

Version history

1.7.3latest on PyPI · released Oct 16, 2025

Audit

Dependencies

numpyrequiredArray operations and mathematical computations.

scipyrequiredSpatial distance calculations and optimization routines.

scikit-learnrequiredFor the sklearn-compatible Kriging interface.

Agent activity

37 hits · last 30 days

node

Resources

githubhttps://github.com/GeoStat-Framework/PyKrige ↗

homepagehttps://github.com/GeoStat-Framework/PyKrige ↗