Documentation

This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English verison of the page.

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

LeastSquaresResults object

Results object containing estimation results from least-squares regression

Description

The LeastSquaresResults object is a superclass of two results objects: NLINResults object and OptimResults object. These objects contain estimation results from fitting a SimBiology® model to data using sbiofit with any supported algorithm.

If sbiofit uses the nlinfit estimation algorithm, the results object is the NLINResults object. If sbiofit uses any other supporting algorithm, then the results object is an OptimResults object. See the sbiofit function for the list of supported algorithms.

Method Summary

boxplot(LeastSquaresResults,OptimResults,NLINResults)Create box plot showing the variation of estimated SimBiology model parameters
fitted(LeastSquaresResults,OptimResults,NLINResults)Return simulation results of SimBiology model fitted using least-squares regression
plot(LeastSquaresResults,OptimResults,NLINResults)Compare simulation results to the training data, creating a time-course subplot for each group
plotActualVersusPredicted(LeastSquaresResults,OptimResults,NLINResults)Compare predictions to actual data, creating a subplot for each response
plotResidualDistribution(LeastSquaresResults,OptimResults,NLINResults)Plot the distribution of the residuals
plotResiduals(LeastSquaresResults,OptimResults,NLINResults)Plot residuals for each response, using time, group, or prediction as x-axis
predict(LeastSquaresResults,OptimResults,NLINResults)Simulate and evaluate fitted SimBiology model
random(LeastSquaresResults,OptimResults,NLINResults)Simulate SimBiology model, adding variations by sampling error model
summary(LeastSquaresResults,OptimResults,NLINResults)Plot a summary figure that contains estimated values and estimation statistics

Properties

GroupNameCategorical variable representing the name of the group associated with the results, or [] if the 'Pooled' name-value pair argument was set to true when you ran sbiofit.
BetaTable of estimated parameters where the jth row represents the jth estimated parameter βj. It contains transformed values of parameter estimates if any parameter transform is specified.

Standard errors of these parameter estimates (StandardError) are calculated as: sqrt(diag(COVB)).

It can also contain the following variables:

  • Bounds — the values of transformed parameter bounds that you specified during fitting

  • CategoryVariableName — the names of categories or groups that you specified during fitting

  • CategoryValue — the values of category variables specified by CategoryVariableName

This table contains one row per distinct parameter value.
ParameterEstimatesTable of estimated parameters where the jth row represents the jth estimated parameter βj. This table contains untransformed values of parameter estimates.

Standard errors of these parameter estimates (StandardError) are calculated as: sqrt(diag(CovarianceMatrix)).

It can also contain the following variables:

  • Bounds — the values of parameter bounds that you specified during fitting

  • CategoryVariableName — the names of categories or groups that you specified during fitting

  • CategoryValue — the values of category variables specified by CategoryVariableName

This table contains sets of parameter values that are identified for each individual or group.
JJacobian matrix of the model, with respect to an estimated parameter, that is,

J(i,j,k)=ykβj|ti

where ti is the ith time point, βj is the jth estimated parameter in the transformed space, and yk is the kth response in the group of data.

COVBEstimated covariance matrix for Beta, which is calculated as: COVB = inv(J'*J)*MSE.
CovarianceMatrixEstimated covariance matrix for ParameterEstimates, which is calculated as: CovarianceMatrix = T'*COVB*T, where T = diag(JInvT(Beta)).

JInvT(Beta) returns a Jacobian matrix of Beta which is inverse transformed accordingly if you specified any transform to estimated parameters.

For instance, suppose you specified the log-transform for an estimated parameter x when you ran sbiofit. The inverse transform is: InvT = exp(x), and its Jacobian is: JInvT = exp(x) since the derivative of exp is also exp.

RResiduals matrix where Rij is the residual for the ith time point and the jth response in the group of data.
LogLikelihoodMaximized loglikelihood for the fitted model.
AICAkaike Information Criterion (AIC), calculated as AIC = 2*(-LogLikelihood + P), where P is the number of parameters.
BICBayes Information Criterion (BIC), calculated as BIC = -2*LogLikelihood + P*log(N), where N is the number of observations, and P is the number of parameters.
DFEDegrees of freedom for error, calculated as DFE = N-P, where N is the number of observations and P is the number of parameters.
MSEMean squared error.
SSESum of squared (weighted) errors or residuals.
WeightsMatrix of weights with one column per response and one row per observation.
EstimatedParameterNamesCell array of character vectors specifying estimated parameter names.
ErrorModelInfoTable describing the error models and estimated error model parameters.

  • It has one row per error model.

  • The ErrorModelInfo.Properties.RowsNames property identifies which responses the row applies to.

  • The table contains three variables: ErrorModel, a, and b. The ErrorModel variable is categorical. The variables a and b can be NaN when they do not apply to a particular error model.

There are four built-in error models. Each model defines the error using a standard mean-zero and unit-variance (Gaussian) variable e, the function value f, and one or two parameters a and b. In SimBiology, the function f represents simulation results from a SimBiology model.

  • 'constant': y=f+ae

  • 'proportional': y=f+b|f|e

  • 'combined': y=f+(a+b|f|)e

  • 'exponential': y=fexp(ae)

EstimationFunctionName of the estimation function.

Note

Loglikelihood, AIC, and BIC properties are empty for LeastSquaresResults objects that were obtained before R2016a.

Introduced in R2014a

Was this topic helpful?