py3dinterpolations.modelling.interpolate.interpolate

py3dinterpolations.modelling.interpolate.interpolate(griddata: GridData, model_name: str, grid_resolution: float, model_params: dict = {}, model_params_grid: dict = {}, preprocess_kwags: dict = {}, predict_kwags: dict = {}, return_model: bool = False) → Union[ndarray, Tuple[ndarray, Modeler]]

interpolate griddata

Interpolate griddata using a Modeler instance that wraps all supported models. The model is selected with the argument model_name.

If the model_params is passed, then the model is initialized with those parameters. Otherwise, to make a search for the best parameters, use the model_params_grid argument.

The 3D grid for interpolation is retrived from the training data. At the moment features only a regular grid.

If requested, the griddata is preprocessed using the Preprocessor class.

Parameters

  • griddata (GridData) – griddata to interpolate

  • model_name (str) – model name

  • grid_resolution (float) – grid resolution

  • model_params (dict, optional) – model parameters. Defaults to {}.

  • preprocess_kwags (dict, optional) – preprocessing parameters. Defaults to {}.

  • predict_kwags (dict, optional) – prediction parameters. Defaults to {}.

  • return_model (bool, optional) – return model. Defaults to False.

Returns

interpolated griddata, optionally with model

Return type

Union[np.ndarray, Tuple[np.ndarray, Modeler]]

Raises

  • ValueError – either model_params or model_params_grid must be passed

  • ValueError – model_params and model_params_grid cannot be passed together

  • NotImplementedError – only RegularGrid3D is supported.

  • NotImplementedError – Parameter search is only supported for ordinary_kriging

Examples

>>> # interpolate griddata
>>> interpolated = interpolate(
>>>     griddata,
>>>     model_name="ordinary_kriging",
>>>     grid_resolution=5,
>>>     model_params={
>>>         "variogram_model": "linear",
>>>         "nlags": 6,
>>>         "weight": True,
>>>     },
>>>     preprocess_kwags={
>>>         "downsampling_res": 0.1,
>>>         "normalize_xyz": True,
>>>         "standardize_v": True,
>>>     },
>>> )