shnitsel.analyze.pls

Functions

pls(xdata_array, ydata_array[, n_components, common_dim])

Performs the partial least squares analysis on the two data arrays provided as arguments and returns the

pls_ds(dataset, xname, yname[, n_components, common_dim])

Wrapper function to perform partial least square analysis on two variables of a dataset

Module Contents

pls(xdata_array, ydata_array, n_components=2, common_dim=None)

Performs the partial least squares analysis on the two data arrays provided as arguments and returns the requested number of resulting main components.

Parameters:

  • xdata_array (xr.DataArray) – First set of data. Shape should be (n_samples, n_features) with n_samples being the length of the common dimension with ydata_array.

  • ydata_array (xr.DataArray) – Second set of data. Shape should be (n_samples, n_targets) with n_samples being the length of the common dimension with xdata_array.

  • n_components (int, optional) – Number of most relevant main components that should be returned. Defaults to 2.

  • common_dim (str | None, optional) – The common dimension which should not be reduced in the course of the analysis. Defaults to None and will attempt to find a single common dimension.

Raises:

  • ValueError – If either xdata_array or ydata_array do not have exactly 2 dimensions.

  • ValueError – If no common_dim was set and x and y data did not have exactly 1 dimension in common to allow for automatic identification of the common dimension.

Returns:

The dataset holding the results of the PLS analysis. Results will either be in variables with the same name as xdata_array or ydata_array or in variables x and y if the names on the respective array are not set.

Return type:

xr.Dataset

pls_ds(dataset, xname, yname, n_components=2, common_dim=None)

Wrapper function to perform partial least square analysis on two variables of a dataset

Parameters:

  • dataset (xr.Dataset) – The dataset holding the variables to apply PLS to

  • xname (str) – The name of the variable to use as the x data for the PLS.

  • yname (str) – The name of the variable to use as the y data for the PLS.

  • n_components (int, optional) – Number of most relevant main components that should be returned. Defaults to 2.

  • common_dim (str | None, optional) – The common dimension which should not be reduced in the course of the analysis. Defaults to None and will attempt to find a single common dimension.

Returns:

The result of the call to pls(). Has the results of the PLS as variables in either the same names as xname and yname or in x and y.s

Return type:

xr.Dataset