[D,FigH,AxH,ObjH,LikH,EstH,BH] = neighbourhood(M,PS,Neigh,...)
M
[ model | bkwmodel ] - Model or bkwmodel object.
PS
[ poster ] - Posterior simulator (poster) object returned by the model/estimate
function.
Neigh
[ numeric ] - The neighbourhood in which the objective function will be evaluated defined as multiples of the parameter estimates.
D
[ struct ] - Struct describing the local behaviour of the objective function and the data likelihood (minus log likelihood) within the specified range around the optimum for each parameter.The following output arguments are non-empty only if you choose 'plot='
true:
FigH
[ numeric ] - Handles to the figures created.
AxH
[ numeric ] - Handles to the axes objects created.
ObjH
[ numeric ] - Handles to the objective function curves plotted.
LikH
[ numeric ] - Handles to the data likelihood curves plotted.
EstH
[ numeric ] - Handles to the actual estimate marks plotted.
BH
[ numeric ] - Handles to the bounds plotted.
'plot='
[ true
| false
] - Call the grfun.plotneigh
function from within neighbourhood
to visualise the results.
'neighbourhood='
[ struct | empty ] - Struct specifying the neighbourhood points for some of the parameters; these points will be used instead of those based on Neigh
.
See help on grfun.plotneigh
for options available when you choose 'plot='
true.
In the output database, D
, each parameter is a 1-by-3 cell array. The first cell is a vector of parameter values at which the local behaviour was investigated. The second cell is a matrix with two column vectors: the values of the overall minimised objective function (as set up in the estimate
function), and the values of the data likelihood component. The third cell is a vector of four numbers: the parameter estimate, the value of the objective function at optimum, the lower bound and the upper bound.