You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the paramGUI interface a user can specify a model to fit the freshly simulated data (#183) in a simple manner. Rather than needing to think about model files and/or parameter files in some kind of yaml specification, following some kind of syntax they can just do the following
select a model type (e.g. kinetic) - required
specify some number of decay rates - required
optionally specify if a gaussian IRF is to be used
optionally specify if backsweep is to be used
specify if the model to be used is parallel or sequential (unbranched unidirectional model )
optionally force positive decay rates
specify the maximum number of iterations (nice to have)
In a notebook this would look like, e.g.:
gta.fit(data, type='kinetic', mode='sequential', rates=(0.055, 0.005), irf=(2.0,1.0), backsweep = True)
(if backsweep is True use default value 13164.8235)
The didactic workflow here is then that the user first if confronted / presented with the Fit progression and from this needs to judge if the fit went well and has converged. Without convergence the fit results always need to be interpreted with a grain of salt, or not at all.
The standard reporting method should show the elements as per the screenshot, e.g. concentrations, spectra, singular values SVD residual, 1st LSV residual, 1st RSV residual, the (raw) residuals, optionally depicting the orignal data (data + (1st, 2nd) SVD).
Screenshots
standard result report
diagnostics
The text was updated successfully, but these errors were encountered:
In the paramGUI interface a user can specify a model to fit the freshly simulated data (#183) in a simple manner. Rather than needing to think about model files and/or parameter files in some kind of yaml specification, following some kind of syntax they can just do the following
In a notebook this would look like, e.g.:
gta.fit(data, type='kinetic', mode='sequential', rates=(0.055, 0.005), irf=(2.0,1.0), backsweep = True)
(if backsweep is True use default value 13164.8235)
The didactic workflow here is then that the user first if confronted / presented with the Fit progression and from this needs to judge if the fit went well and has converged. Without convergence the fit results always need to be interpreted with a grain of salt, or not at all.
The standard reporting method should show the elements as per the screenshot, e.g. concentrations, spectra, singular values SVD residual, 1st LSV residual, 1st RSV residual, the (raw) residuals, optionally depicting the orignal data (data + (1st, 2nd) SVD).
Screenshots
standard result report
diagnostics
The text was updated successfully, but these errors were encountered: