OPLS

Functionality for working with Orthogonal Partial Least Squares (OPLS) modeling.

class friendly_mvda.models.opls_model.FittedOplsModel(model: OplsModel, model_id: int)

Bases: object

property DmodX: DataFrame: Distance to the model in X space

property DmodX_limit: DataFrame: get the critical limits of the score. Only works with evolution models.

property Hotellings_T2: DataFrame: Hotelling’s T2 from the model

property Q2: DataFrame: Cumulative predicted fraction of the variation of the Y-variables, according to cross-validation, get data for all components.

property R2: DataFrame: Cumulative explained fraction of the variation, get data for all components.

property RMSECV: DataFrame: Root Mean Square Error, computed from the selected cross-validation round.

property RMSEE: DataFrame: Root Mean Square Error of the Estimation.

property aligned_variable_values: DataFrame: Get the aligned variables from the model. only works with evolution models.

property batch_limits: DataFrame: Get all limits with aligned batch length

contributions(batch: str, time: float, contribution_type: ContributionType) → ndarray[tuple[int, ...], dtype[Any]]

property loadings: DataFrame: Loadings from the model

property loadings_correlation_scaled: DataFrame: Loadings(correlation scaled) from the model

property number_of_components: list[int]: Get the number of components used

property observed_y: DataFrame: Observed values of Y-variables for observations in the model.

classmethod open_model(path: str, model_id: int = 1) → FittedOplsModel: Open an existing fitted model.

predict(data: DataFrame) → DataFrame

Use the model to get the prediction of the observation.

use the name of the variables to identify them.

Parameters

data : The observation to get the prediction of.

Returns

y_pred : The prediction of the observations.

RMSEP : The root mean square error of the predictions, require data to contains y data.

property predicted_y: DataFrame: Predicted values of Y-variables for observations in the model.

refit(predictive_components: int = -1, orthogonal_components: int = -1, orthogonal_y_components: int = -1) → None: Refit the model with another number of components. The previous fitted model will not work anymore.

save(path: str) → None

Save the model.

parameters:

path: the path to the saved file.

property score_limits: DataFrame: Get the critical limits of the score. Only works with evolution models.

property scores: DataFrame: Scores from the model

set_limit_confidence_level(confidence_level: float) → None: Set the confidence level of the limits

property variable_limits: DataFrame: Get the critical limits of the X variables. only works with evolution models.

class friendly_mvda.models.opls_model.OplsModel(project: Project)

Bases: object

change_missing_tolerance(tolerance: float) → None

Change the tolerance for when data will be excluded because of missing value any variable of an observation with more missing values than the tolerance will be excluded.

Default is 50(%)

change_scaling(scal_type: str, variable: str | None = None) → None: Change the scaling of the model

fit(predictive_components: int = -1, orthogonal_components: int = -1, orthogonal_y_components: int = -1) → FittedOplsModel

Fit the model.

parameters:

predictive_components: the number of predictive components, -1 is auto fit

orthogonal_components: the number of orthogonal components, -1 is auto fit

orthogonal_y_component: the number of orthogonal y components, -1 is auto fit

returns:

FittedOplsModel: the fitted opls model.

example:

>>> from friendly_mvda.models.opls_model import OplsModel
>>> model = OplsModel(project)
>>> model.remove_variables(["pH"])
>>> model.remove_observation([46, 92])
>>> model.change_scaling("uv")
>>> model.change_missing_tolerance(30)
>>> fitted_model = model.fit(2, 1, 0)

here the model is modified before fitting by removing variable “pH”, removing observations 46 and 92, changing scaling to uv and changing missing tolerance to 30 %. Finally the model is fitted with 2 predictive components and 1 orthogonal component.

classmethod open_model(path: str, model_id: int = 1) → OplsModel

Open a existing model.

parameters:

path: the path to the saved file.

model_id: the index of the model. Can be used if the file contains multiple models.

remove_batches(batches: list[str]) → None: Remove batches from the model

remove_observation(observations: list[int]) → None: Remove observations from the model

remove_variables(variables: list[str]) → None: Remove variables from the model

save(path: str) → None

Save the model

parameters:

path: the path to the saved file.