ModelFitResult¶

class fitr.inference.modelfitresult.MCMCFitResult(method, nsubjects, nparams, name)¶

Results of model fitting with MCMC

Attributes:

name : str: Model identifier. We suggest using free-parameters as identifiers
method : str: Method employed in optimization.
nsubjects : int: Number of subjects fitted.
nparams : int: Number of free parameters in the fitted model.
params : ndarray(shape=(nsubjects, nparams)): Array of parameter estimates
paramnames : list: List of parameter names
stanfit :: Stan fit object
summary : pandas.DataFrame: Summary of the MCMC fit results

Methods

get_paramestimates(self, FUN=np.mean)	Extracts parameter estimates
trace_plot(self, figsize=None, save_figure=False, filename=’fitr-mcstan-traceplot.pdf’)	Trace plot for fit results

get_paramestimates(FUN=<Mock name='mock.median' id='140316081412640'>)¶

Extracts parameter estimates

Parameters:	FUN : {numpy.mean, numpy.median}

trace_plot(figsize=None, save_figure=False, filename='fitr-mcstan-traceplot.pdf')¶

Easy wrapper for Stan Traceplot

Parameters:	figsize : (optional) list [width in inches, height in inches] Controls figure size save_figure : bool Whether to save the figure to disk filename : str The file name to be output

class fitr.inference.modelfitresult.ModelFitResult(method, nsubjects, nparams, name=None)¶

Class representing the results of a fitrmodel fitting.

Attributes:	name : str Model identifier. We suggest using free-parameters as identifiers method : str Method employed in optimization. nsubjects : int Number of subjects fitted. nparams : int Number of free parameters in the fitted model. params : ndarray(shape=(nsubjects, nparams)) Array of parameter estimates paramnames : list List of parameter names

Methods

set_paramnames(params)	Sets names of RL parameters to the fitrfit object
plot_ae(actual, save_figure=False, filename=’actual-estimate.pdf’)	Plots estimated parameters against actual simulated parameters
summary_table(write_csv=False, filename=’summary-table.csv’, delimiter=’,’)	Writes a CSV file with summary statistics from the present model

ae_metrics(actual, matches=None)¶

Computes metrics summarizing the ability of the model to fit data generated from a known model

Parameters:	matches : list List consisting of [rlparams object, column index in actual, column index in estimates]. Ensures comparisons are being made between the same parameters, particularly when the models have different numbers of free parameters.
Returns:	DataFrame Including summary statistics of the parameter matching

plot_ae(actual, save_figure=False, filename='actual-estimate.pdf')¶

Plots actual parameters (if provided) against estimates

Parameters:	actual : ndarray(shape=(nsubjects, nparams)) Array of actual parameters from a simulation save_figure : bool Whether to save the figure to disk filename : str The file name to be output

set_paramnames(params)¶

Sets the names of the RL parameters to the fitrfit object

Parameters:	params : list List of parameters from the rlparams module

class fitr.inference.modelfitresult.OptimizationFitResult(method, nsubjects, nparams, name)¶

Results of model fitting with optimization methods

Attributes:

name : str: Model identifier. We suggest using free-parameters as identifiers
method : str: Method employed in optimization.
nsubjects : int: Number of subjects fitted.
nparams : int: Number of free parameters in the fitted model.
params : ndarray(shape=(nsubjects, nparams)): Array of parameter estimates
paramnames : list: List of parameter names
errs : ndarray(shape=(nsubjects, nparams)): Array of parameter estimate errors
nlogpost : ndarray(shape=(nsubjects)): Subject level negative log-posterior probability
nloglik : float: Subject level negative log-likelihood
LME : float: Log-model evidence
BIC : ndarray(shape=(nsubjects)): Subject-level Bayesian Information Criterion
AIC : ndarray(shape=(nsubjects)): Subject-level Aikake Information Criterion
summary : DataFrame: Summary of means and standard deviations for each free parameter, as well as negative log-likelihood, log-model-evidence, BIC, and AIC for the model

Methods

plot_fit_ts(save_figure=False, filename=’fit-stats.pdf’) :	Plots the evolution of log-likelihood, log-model-evidence, AIC, and BIC over optimization iterations
param_hist(save_figure=False, filename=’param-hist.pdf’) :	Plots hitograms of parameters in the model
summary_table(write_csv=False, filename=’summary-table.csv’, delimiter=’,’)	Writes a CSV file with summary statistics from the present model

param_hist(save_figure=False, filename='param-hist.pdf')¶

Plots histograms of the parameter estimates

Parameters:	save_figure : bool Whether to save the figure to disk filename : str The file name to be output

plot_fit_ts(save_figure=False, filename='fit-stats.pdf')¶

Plots the log-model-evidence, BIC, and AIC over optimization iterations

Parameters:	save_figure : bool Whether to save the figure to disk filename : str The file name to be output