deepsensor 0.1.2__tar.gz → 0.1.3__tar.gz

{deepsensor-0.1.2 → deepsensor-0.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepsensor
-Version: 0.1.2
+Version: 0.1.3
 Summary: A Python package for modelling xarray and pandas data with neural processes.
 Home-page: https://github.com/tom-andersson/deepsensor
 Author: Tom R. Andersson

{deepsensor-0.1.2 → deepsensor-0.1.3}/deepsensor.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepsensor
-Version: 0.1.2
+Version: 0.1.3
 Summary: A Python package for modelling xarray and pandas data with neural processes.
 Home-page: https://github.com/tom-andersson/deepsensor
 Author: Tom R. Andersson

{deepsensor-0.1.2 → deepsensor-0.1.3}/deepsensor.egg-info/SOURCES.txt

@@ -1,22 +1,17 @@
-LICENSE
 README.md
 pyproject.toml
 setup.cfg
 setup.py
 deepsensor/__init__.py
-deepsensor/_version.py
 deepsensor/config.py
 deepsensor/plot.py
 deepsensor/py.typed
-deepsensor/utils.py
 deepsensor.egg-info/PKG-INFO
 deepsensor.egg-info/SOURCES.txt
 deepsensor.egg-info/dependency_links.txt
 deepsensor.egg-info/not-zip-safe
 deepsensor.egg-info/requires.txt
 deepsensor.egg-info/top_level.txt
-deepsensor/active_learning/__init__.py
-deepsensor/active_learning/acquisition_fns.py
 deepsensor/data/__init__.py
 deepsensor/data/loader.py
 deepsensor/data/processor.py

{deepsensor-0.1.2 → deepsensor-0.1.3}/deepsensor.egg-info/requires.txt

@@ -8,6 +8,7 @@ dask
 distributed
 pyshp
 tqdm
+pooch
 
 [testing]
 pytest

{deepsensor-0.1.2 → deepsensor-0.1.3}/pyproject.toml

@@ -4,10 +4,6 @@ testpaths = [
     "tests",
 ]
 
-
-[tool.setuptools_scm]
-write_to = "deepsensor/_version.py"
-
 [tool.mypy]
 mypy_path = "deepsensor"
 check_untyped_defs = true

{deepsensor-0.1.2 → deepsensor-0.1.3}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = deepsensor
-version = 0.1.2
+version = 0.1.3
 author = Tom R. Andersson
 author_email = tomand@bas.ac.uk
 description = A Python package for modelling xarray and pandas data with neural processes.
@@ -31,6 +31,7 @@ install_requires =
 	distributed
 	pyshp
 	tqdm
+	pooch
 
 [options.extras_require]
 testing = 

deepsensor-0.1.2/LICENSE

@@ -1,21 +0,0 @@
-MIT License
-
-Copyright (c) 2023 Tom Robin Andersson
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

deepsensor-0.1.2/deepsensor/_version.py

@@ -1,12 +0,0 @@
-import os
-
-
-def string():
-    try:
-        with open(os.path.dirname(__file__) + "/VERSION", "r", encoding="utf-8") as fh:
-            version = fh.read().strip()
-            if version:
-                return version
-    except:
-        pass
-    return "unknown (git checkout)"

deepsensor-0.1.2/deepsensor/active_learning/acquisition_fns.py

@@ -1,220 +0,0 @@
-class GreedySensorSearch:
-    """
-    Parent class for a greedy search algorithm that proposes locations for
-    new sensor measurements from a starting set of query locations.
-    """
-
-    def __init__(
-        self,
-        tasks,
-        N_new_sensors,
-        region,
-        data_processor_ID,
-        reference_grid=None,
-        mask_da=None,
-        context_set_idx=None,
-        target_var=None,
-        maxormin="max",
-        progressbar=0,
-    ):
-        """
-        Args:
-            tasks (dict): Dictionary mapping from dates of observation to
-                tasks output by child of `ToySensorData`.
-                For each task:
-                    - The `X_target` entry is treated as the initial search space
-                        (assumed the same for each task).
-                    - The `X_context`, `Y_context` entries comprise the context set
-                        (starting sensor network).
-
-            N_new_sensors (int): The total number of new sensor locations to propose.
-
-            reference_grid (xr.DataArray):
-                If running search on a grid, contains the grid coordinates
-                to search over. Used for instantiating an xr.DataArray storing acquisition
-                function values.
-
-            mask_da (boolean xr.DataArray):
-                If running search on a grid, contains the grid coordinates to remove from the
-                search (example use case: masking out the ocean for land sensors). Boolean True
-                values are interpreted as valid points, and False as invalid points. The mask will
-                be interpolated onto the search grid using nearest-neighbour interpolation
-                (i.e. the mask and search grid do not need to be the same).
-
-            maxormin (str): 'max' or 'min' - whether to maximise or minimise the
-                acquisition function within each greedy search iteration.
-
-            progressbar (int): Progress bar verbosity:
-                -   0-2: no effect
-                -   3: sensor progressbar
-                -   4: date progressbar
-                -   5: search progressbar
-        """
-        self.tasks = tasks
-        self.N_new_sensors = N_new_sensors
-        self.region = region
-        self.data_processor_ID = data_processor_ID
-        self.reference_grid = reference_grid
-        self.mask_da = mask_da
-        self.context_set_idx = context_set_idx
-        self.maxormin = maxormin
-        self.progressbar = progressbar
-
-        # TODO use to determine target_set_idx?
-        self.target_var = target_var
-
-        self.dates = list(self.tasks.keys())
-
-        # Don't edit by reference if running multiple classes with same `tasks` input
-        self.tasks = copy.deepcopy(self.tasks)
-
-        # Uninstrumented locations to search over: initialise as whole search space
-        self.X_search_init = self.tasks[self.dates[0]]["X_search"][0].copy()
-        if self.X_search_init.ndim == 3:
-            self.X_search_init = self.X_search_init[0]
-        self.search_size = self.X_search_init.shape[-1]
-        self.check_X_search_large_enough()
-
-    def check_X_search_large_enough(self):
-        """Ensure number of sensors to place is less than the search space size."""
-        if self.search_size < self.N_new_sensors:
-            raise ValueError(
-                f"Initial search size ({self.search_size}) is smaller "
-                f"than number of sensors to place ({self.N_new_sensors})."
-            )
-        elif self.search_size == self.N_new_sensors:
-            raise ValueError(
-                f"Initial search size ({self.search_size}) is equal to "
-                f"the number of sensors to place ({self.N_new_sensors})."
-            )
-        else:
-            print(
-                f"Placing {self.N_new_sensors} sensors in search space of {self.search_size}.\n"
-            )
-
-    def acquisition_fn(self, task, x_query, sample_i):
-        """
-        Acquisition function for assigning an importance value to a single
-        sensor location `x_query`. To be overridden by child classes implementing
-        specific acquisition functions for the greedy search.
-
-        Child class overridden methods should take the form:
-            ```
-            task = self.append_query_obs_to_task(task, x_query, sample_i)
-            task = self.sample_task(task, sample_i)
-            # Code to compute importance of `x_query...
-            return importance
-            ```
-        """
-        raise NotImplementedError()
-
-    def search(self, X_search):
-        """
-        Run one greedy pass by looping over each point in `X_search` and
-        computing the acquisition function.
-
-        If the search algorithm can be run over all points in parallel,
-        this method should be overridden by the child class so that `self.run()`
-        uses the parallel implementation.
-
-        TODO check if below is still valid in GreedyOptimal:
-        If the search method uses the y-values at search points (i.e. for
-        an optimal benchmark), its `acquisition_fn` should expect a y_query input.
-        """
-        importances_list = []
-        for date, task in tqdm(
-            self.tasks.items(),
-            desc="Date",
-            position=4,
-            leave=True,
-            disable=self.progressbar < 4,
-        ):
-            for sample_i in range(self.n_samples_or_1):
-                # Add size-1 dim after row dim to preserve row dim for passing to
-                #   acquisition_fn. Also roll final axis to first axis for looping over search points.
-                importances = []
-                for x_query in tqdm(
-                    np.rollaxis(X_search[:, np.newaxis], 2),
-                    desc="Search",
-                    position=6,
-                    leave=True,
-                    disable=self.progressbar < 5,
-                ):
-                    importance = self.acquisition_fn(task, x_query, sample_i)
-                    importances.append(importance)
-
-                importances = np.array(importances)
-                self.store_acquisition_fn_values(
-                    self.iteration, date, sample_i, importances
-                )
-                importances_list.append(importances)
-
-        return np.mean(importances_list, axis=0)
-
-    def select_best(self, importances, X_search):
-        """Select sensor location corresponding to best importance value.
-
-        Appends the chosen search index to a list of chosen search indexes.
-        """
-        if self.maxormin == "max":
-            best_idx = np.argmax(importances)
-        elif self.maxormin == "min":
-            best_idx = np.argmin(importances)
-
-        best_x_query = X_search[:, best_idx : best_idx + 1]
-
-        # Index into original search space of chosen sensor location
-        self.best_idxs_all.append(
-            np.where((self.X_search_init == best_x_query).all(axis=0))[0][0]
-        )
-
-        return best_x_query
-
-    def single_greedy_iteration(self):
-        """
-        Run a single greedy grid search iteration and append the optimal
-        sensor location to self.X_new.
-        """
-        importances = self.search(self.X_search)
-        best_x_query = self.select_best(importances, self.X_search)
-
-        # Remove greedily chosen sensor from search space
-        # TEMP turning off this so that results always on grid for storing in xarray
-        #   This means 'waste' predictions over previously placed sensors/risk placing
-        #   multiple sensors at same location!
-        # self.X_search = self.X_search[~(self.X_search == best_x_query).all(axis=1)]
-
-        self.X_new.append(best_x_query)
-        self.iteration += 1
-
-    def run(self):
-        """
-        Iteratively... docstring TODO
-
-        Returns a tensor of proposed new sensor locations (in greedy iteration/priority order)
-            and their corresponding list of indexes in the search space.
-        """
-        # Have separate `self.X_search` in case search space changes as placements are made
-        #   (although this functionality isn't implemented yet)
-        self.X_search = self.X_search_init
-        self.iteration = 0
-
-        # List of new proposed sensor locations
-        self.X_new = []
-
-        # List to track indexes into original search grid of chosen sensor locations
-        #   as optimisation progresses. Used for filling y-values at chosen
-        #   sensor locations, `self.X_new`
-        self.best_idxs_all = []
-
-        for _ in tqdm(
-            range(self.N_new_sensors),
-            desc="Placement",
-            position=2,
-            leave=True,
-            disable=self.progressbar < 3,
-        ):
-            self.single_greedy_iteration()
-
-        X_new = np.concatenate(self.X_new, axis=1)
-        return X_new, self.best_idxs_all