formula

ConditionalRegressionMixin.formula(labels=None)

Creates the prediction formula of the fit for the encoded and normalized data. The labels for each term are given by the labels parameter. The number of labels must be the number of parameters from Y2X, i.e., len(labels) == Y2X(Y).shape[1].

It also includes the conditional effects automatically.

Warning

This formula is the prediction formula of the encoded and normalized data. First apply factor normalization and then categorical encoding before applying this prediction formula.

>>> # Imports
>>> from numba.typed import List
>>> from pyoptex.utils import Factor
>>> from pyoptex.utils.design import encode_design
>>> 
>>> # Example factors
>>> factors = [
>>>     Factor('A'), 
>>>     Factor('B'),
>>>     Factor('C', type='categorical', levels=['L1', 'L2', 'L3'])
>>> ]
>>> 
>>> # Compute derived parameters
>>> effect_types = np.array([
>>>     1 if f.is_continuous else len(f.levels)
>>>     for f in factors
>>> ])
>>> coords = List([f.coords_ for f in factors])
>>> 
>>> # Normalize the factors
>>> for f in factors:
>>>     data[str(f.name)] = f.normalize(data[str(f.name)])
>>> 
>>> # Select correct order + to numpy
>>> data = data[[str(f.name) for f in factors]].to_numpy()
>>> 
>>> # Encode
>>> data = encode_design(data, effect_types, coords=coords)
>>> 
>>> # Transform according to the model
>>> data = Y2X(data)

Note

If you created Y2X using model2Y2X, use model_formula <pyoptex.analysis.mixins.fit_mixin.model_formula(). It will automatically assign the correct labels.

Parameters

labelslist(str)

The list of labels for each encoded, normalized term.

Returns

formulastr

The prediction formula for encoded and normalized data.