Add dataworth plugin

docs/examples/index.rst

@@ -45,6 +45,13 @@ Below you can find examples for the different plugins.
 
     randomized_maximum_likelihood
 
+.. toctree::
+    :maxdepth: 4
+    :hidden:
+    :caption: Data Worth
+
+    randomized_maximum_likelihood
+
 
 Response functions
 ------------------
@@ -87,4 +94,10 @@ PEST Solver
 Randomized Maximum Likelihood
 -----------------------------
 
-:doc:`Randomized Maximum Likelihood parameter estimation <randomized_maximum_likelihood>`
+:doc:`Randomized Maximum Likelihood parameter estimation <randomized_maximum_likelihood>`
+
+
+Data Worth
+----------
+
+:doc:`Data Worth <dataworth>`

docs/plugins/dataworth.rst

@@ -0,0 +1,46 @@
+.. _DataWorth:
+
+Data Worth
+----------
+
+.. role:: python(code)
+   :language: python
+
+The :code:`dataworth` plugin contains a class and and visualizations to analyze the 
+data worth of observations in a pastas model.
+
+Currently the following class is available:
+
+- :class:`~pastas_plugins.dataworth.DataWorth`: Create a DataWorth class to run data worth analysis
+
+This class exposes the following methods to analyze data worth:
+
+- :meth:`~pastas_plugins.dataworth.DataWorth.data_worth_per_observation`: Compute data worth per observation with leave-one-out analysis.
+- :meth:`~pastas_plugins.dataworth.DataWorth.data_worth_thinning`: Compute data worth for groups of observations by specifying thinning factors.
+- :meth:`~pastas_plugins.dataworth.DataWorth.data_worth_per_added_observation`: Compute data worth for each added observation.
+- :meth:`~pastas_plugins.dataworth.DataWorth.data_worth_new_observations`: Compute data worth for groups of added observations.
+- :meth:`~pastas_plugins.dataworth.DataWorth.recompute_jacobian`: Recompute jacobian, needed to include new observations for data worth analysis.
+
+These methods are the backbone of the above calculations:
+
+- :meth:`~pastas_plugins.dataworth.DataWorth.observation_noise_covariance`: Compute an observation noise covariance matrix.
+- :meth:`~pastas_plugins.dataworth.DataWorth.fisher_information_matrix`: Compute the Fisher information matrix.
+- :meth:`~pastas_plugins.dataworth.DataWorth.compute_covariance`: Compute parameter covariance matrix, given a jacobian, an observation noise covariance matrix and an optional mask
+- :meth:`~pastas_plugins.dataworth.DataWorth.data_worth`: Compute overall data worth (log determinant of parameter covariance matrix) and per-parameter data worth (variance of parameters).
+
+The following plots are available:
+
+- :func:`~pastas_plugins.dataworth.plot_data_worth_series`: Plot the overall data worth and per parameter data worth for each per observation.
+- :func:`~pastas_plugins.dataworth.plot_data_worth_heatmap`: Plot a heatmap of data worth with years on the y-axis and days on the x-axis.
+
+Example
+^^^^^^^
+
+See the :ref:`Examples` section for more information on how to use the :code:`dataworth` plugin.
+
+API
+^^^
+.. automodule:: pastas_plugins.dataworth
+   :members:
+   :undoc-members:
+   :private-members:

pastas_plugins/dataworth/__init__.py

@@ -0,0 +1,7 @@
+# ruff: noqa: F401
+from pastas_plugins.dataworth.dataworth import (
+    DataWorth,
+    plot_data_worth_heatmap,
+    plot_data_worth_series,
+)
+from pastas_plugins.dataworth.version import __version__