predict.PredictiveModelList {lulcc}

Predict location suitability

Description

Estimate location suitability with predictive models.

Usage

## S3 method for class 'PredictiveModelList'
predict(object, newdata, data.frame = FALSE,
  ...)

## S4 method for signature 'PredictiveModelList'
predict(object, newdata, data.frame = FALSE,
  ...)

Arguments

object

a PredictiveModelList object

newdata

data.frame containing new data

data.frame

logical indicating whether the function should return a matrix (default) or data.frame

...

additional arguments to predict methods

Details

This function is usually called from allocate to calculate land use suitability at each timestep. However, it may also be used to produce suitability maps (see examples).

Value

A matrix.

See Also

Model fitting, allocate

Examples

## Not run: 

## Sibuyan Island

## load observed land use data
obs <- ObsLulcRasterStack(x=sibuyan$maps,
                    pattern="lu",
                    categories=c(1,2,3,4,5),
                    labels=c("Forest","Coconut","Grass","Rice","Other"),
                    t=c(0,14))

## load explanatory variables
ef <- ExpVarRasterList(x=sibuyan$maps, pattern="ef")

## separate data into training and testing partitions
part <- partition(x=obs[[1]], size=0.1, spatial=TRUE)
train.data <- getPredictiveModelInputData(obs=obs, ef=ef, cells=part[["train"]])
all.data <- getPredictiveModelInputData(obs=obs, ef=ef, cells=part[["all"]])

## get glm.models from data
forms <- list(Forest ~ ef_001+ef_002+ef_003+ef_004+ef_005+ef_006+ef_007+ef_008+ef_010+ef_012,
              Coconut ~ ef_001+ef_002+ef_005+ef_007+ef_008+ef_009+ef_010+ef_011+ef_012,
              Grass~ef_001+ef_002+ef_004+ef_005+ef_007+ef_008+ef_009+ef_010+ef_011+ef_012+ef_013,
              Rice~ef_009+ef_010+ef_011,
              Other~1)

glm.models <- glmModels(formula=forms, family=binomial, data=train.data, obs=obs)

## create suitability maps
suitability.maps <- predict(object=glm.models, newdata=all.data, data.frame=TRUE)
points <- rasterToPoints(obs[[1]], spatial=TRUE)
suitability.maps <- SpatialPointsDataFrame(coords=points, data=suitability.maps)
r <- stack(rasterize(x=suitability.maps, y=obs[[1]], field=names(suitability.maps)))
plot(r)

## library(rasterVis)
## levelplot(r)


## End(Not run)