================== Migrating to 0.7 ================== Since braincoder 0.7, a model object holds only its configuration: the tuning function, the stimulus grid, the HRF. It never stores data, parameters, weights or a noise model, so nothing you did earlier can change what a later call computes. Passing a removed argument to a constructor raises a ``TypeError`` that points here, and leaving out required inputs raises a ``ValueError``, so old code fails loudly. In short ======== .. list-table:: :header-rows: 1 :widths: 50 50 * - Before 0.7 - Since 0.7 * - ``model = GaussianPRF(paradigm=x, parameters=p)`` ``model.predict()`` - ``model = GaussianPRF()`` ``model.predict(x, p)`` * - ``model.simulate(noise=.1)`` - ``model.simulate(x, p, noise=.1)`` * - ``GaussianPRF2D(grid_coordinates=g, paradigm=x)`` - ``GaussianPRF2D(grid_coordinates=g)`` * - ``GaussianPRFWithHRF(paradigm=x, hrf_model=h)`` - ``GaussianPRFWithHRF(hrf_model=h)`` * - ``RegressionGaussianPRF(paradigm=x, regressors=r)`` - ``RegressionGaussianPRF(design_data=x, regressors=r)`` * - ``model.init_pseudoWWT(grid, p)`` ``ResidualFitter(model, data, x, parameters=p).fit()`` - ``ResidualFitter(model, data, x, parameters=p).fit(stimulus_range=grid)`` * - ``model.get_pseudoWWT()`` - ``model.get_pseudoWWT(grid, p)`` * - ``model.weights = w`` ``model.get_WWT()`` - ``model.get_WWT(w)`` * - ``model.get_stimulus_pdf(data, grid)`` - ``model.get_stimulus_pdf(data, grid, parameters=p, omega=omega, dof=dof)`` * - ``model.get_rf()`` - ``model.get_rf(p)`` * - ``linear = model.to_linear_model()`` - ``linear, baseline, weights = model.to_linear_model(p)`` * - ``xy_model = model.to_xy_model()`` - ``xy_model, xy_parameters = model.to_xy_model(p)`` * - ``CustomStimulusFitter(data, model, stimulus, omega)`` - ``CustomStimulusFitter(data, model, stimulus, omega, parameters=p, weights=w)`` * - ``SPMHRFModel(tr=1.6, unique_hrfs=True)`` - ``SPMHRFModel(tr=1.6)`` * - ``model.apply_mask(mask)`` - Select the voxels in your own ``data``, ``parameters`` and ``weights``. ``ParameterFitter(model, data, paradigm)`` and ``WeightFitter(model, parameters, data, paradigm)`` are unchanged. ``ResidualFitter`` and ``StimulusFitter`` keep the same arguments, but ``parameters`` is now required for any model that has parameters. New fitting default =================== ``ParameterFitter.fit`` now defaults to ``r2_atol=1e-4``, so fits stop earlier. To reproduce a fit from an earlier version, pass ``r2_atol=1e-6``. Custom models ============= Subclasses need two changes: * Remove ``paradigm``, ``data``, ``parameters``, ``weights`` and ``omega`` from ``__init__``, and from the call to ``super().__init__``. * Replace ``self.parameters``, ``self.weights``, ``self.omega`` and ``self.paradigm`` with method arguments. ``_basis_predictions(paradigm, parameters)`` and ``transformations`` work as before. A subclass whose paradigm has more than one stimulus dimension and that uses ``StimulusFitter`` should pass ``n_dimensions`` to the base constructor.