local.models {plspm} | R Documentation |
Calculates PLS-PM for global and local models from a given partition.
local.models(pls, y, scheme=NULL, scaled=NULL, boot.val=FALSE, br=NULL)
pls |
An object of class "plspm" |
y |
One object of the following classes: "rebus" , "integer" , or "factor" , that provides the class partitions. |
scheme |
Possible values are "centroid" , "factor" and "path" . |
scaled |
A logical value indicating whether scaling data is performed. |
boot.val |
A logical value indicating whether bootstrap validation is performed (FALSE by default). |
br |
An integer indicating the number bootstrap resamples. Used only when boot.val=TRUE . |
The function local.models
calculates PLS-PM for the global model (i.e. over all observations) as well as PLS-PM for local models (i.e. observations of different partitions).
When y
is an object of class "rebus"
, the function local.models
is applied to the classes obtained from the REBUS algorithm.
When y
is an integer
vector or a factor
, the values or levels are assumed to represent the group to which each observation belongs. In this case, the function local.models
calculates PLS-PM for the global model, as well as PLS-PM for each group (local models).
If scheme=NULL
, then the original scheme from the object pls
will be taken.
If scaled=NULL
, then the original scaled from the object pls
will be taken.
When bootstrap validation is performed, the default number of re-samples is 200. However, br
can be specified in a range from 50 to 500.
An object of class "local.models"
, basically a list of length k+1
, where k
is the number of classes. The list contains the following elements:
glob.model |
PLS-PM of the global model |
loc.model.1 |
PLS-PM of segment (class) 1 |
loc.model.2 |
PLS-PM of segment (class) 2 |
loc.model.k |
PLS-PM of segment (class) k |
Each element of the list is an object of class "plspm"
. Thus, in order to examine the
results for each local model, it is necessary to use the summary
function. See examples
below.
Note that if scheme
and/or scaled
differ from the original arguments of pls
, the results obtained in local.models
will be different from those obtained in pls
as well as in y
(if it is of class "rebus"
).
Laura Trinchera, Gaston Sanchez
## Not run: ## example of rebus analysis data(sim.data) ## First compute GLOBAL model sim.mat <- matrix(c(0,0,0,0,0,0,1,1,0),3,3,byrow=TRUE) dimnames(sim.mat) <- list(c("Price","Quality","Satisfaction"), c("Price","Quality","Satisfaction")) sim.sets <- list(c(1,2,3,4,5),c(6,7,8,9,10),c(11,12,13)) sim.mod <- c("A","A","A") ## reflective indicators sim.global <- plspm(sim.data, inner.mat=sim.mat, sets=sim.sets, modes=sim.mod) sim.global ## Then compute cluster on residual from global model sim.res.clus <- res.clus(sim.global) ## To conclude run iteration algorithm rebus.sim <- it.reb(sim.global, sim.res.clus, nk=2, stop.crit = 0.005, iter.max = 100 ) ## Computation of local models local.rebus <- local.models(sim.global, rebus.sim) ## Display plspm summary for first local model summary(local.rebus$loc.model.1) ## End(Not run)