#Install necessary packages.

install.packages("dismo", dependencies=TRUE)



#Update packages.

#update.packages("dismo")



#Set the working directory.

setwd("C:/Users/Steve/Desktop/biomod2/biomod2_results/")



#Load package.

library(dismo)



#Load the training data.

DataSpecies_lot_line_section_with_predictors<-read.table(
"C:/Users/Steve/Desktop/biomod2/biomod2_files_to_be_loaded/lot_line_section_with_predictors.csv",
sep=",",
header=TRUE,
colClasses = "numeric")



#Attach the training data.

attach(DataSpecies_lot_line_section_with_predictors)



#Specify the species being modeled.

myRespName.abi.bal <- 'abi.bal'
myRespName.ace.rub <- 'ace.rub'
myRespName.ace.sac <- 'ace.sac'
myRespName.aln.inc <- 'aln.inc'
myRespName.bet.all <- 'bet.all'
myRespName.car.spp <- 'car.spp'
myRespName.cas.den <- 'cas.den'
myRespName.fag.gra <- 'fag.gra'
myRespName.fra.ame <- 'fra.ame'
myRespName.fra.nig <- 'fra.nig'
myRespName.jug.cin <- 'jug.cin'
myRespName.jug.nig <- 'jug.nig'
myRespName.lir.tul <- 'lir.tul'
myRespName.mag.acu <- 'mag.acu'
myRespName.ost.vir <- 'ost.vir'
myRespName.pin.str <- 'pin.str'
myRespName.pla.occ <- 'pla.occ'
myRespName.pru.ser <- 'pru.ser'
myRespName.que.alb <- 'que.alb'
myRespName.que.vel <- 'que.vel'
myRespName.til.ame <- 'til.ame'
myRespName.tsu.can <- 'tsu.can'
myRespName.ulm.ame <- 'ulm.ame'



#Specify the training data.

myData <- DataSpecies_lot_line_section_with_predictors



#Specify the tree complexity.

myTreeComplexity = 5



#Specify which columns in the training data contain the predictor
#variables.  49:62 = includes Native American variables, whereas
#49:59 = excludes Native American variables.

myExpl = 49:62



#Train models to have the optimal number of regression trees.
#See 'dismo' documentation for more details.

myGbmModelOut.abi.bal <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.abi.bal,
family = "bernoulli",
site.weights = abi.bal.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0015,
bag.fraction = 0.5)

myGbmModelOut.ace.rub <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.ace.rub,
family = "bernoulli",
site.weights = ace.rub.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.ace.sac <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.ace.sac,
family = "bernoulli",
site.weights = ace.sac.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.03,
bag.fraction = 0.5)

myGbmModelOut.aln.inc <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.aln.inc,
family = "bernoulli",
site.weights = aln.inc.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.bet.all <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.bet.all,
family = "bernoulli",
site.weights = bet.all.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.car.spp <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.car.spp,
family = "bernoulli",
site.weights = car.spp.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.cas.den <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.cas.den,
family = "bernoulli",
site.weights = cas.den.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0075,
bag.fraction = 0.5)

myGbmModelOut.fag.gra <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.fag.gra,
family = "bernoulli",
site.weights = fag.gra.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.01,
bag.fraction = 0.5)

myGbmModelOut.fra.ame <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.fra.ame,
family = "bernoulli",
site.weights = fra.ame.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.01,
bag.fraction = 0.5)

myGbmModelOut.fra.nig <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.fra.nig,
family = "bernoulli",
site.weights = fra.nig.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.004,
bag.fraction = 0.5)

myGbmModelOut.jug.cin <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.jug.cin,
family = "bernoulli",
site.weights = jug.cin.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.005,
bag.fraction = 0.5)

myGbmModelOut.jug.nig <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.jug.nig,
family = "bernoulli",
site.weights = jug.nig.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.001,
bag.fraction = 0.5)

myGbmModelOut.lir.tul <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.lir.tul,
family = "bernoulli",
site.weights = lir.tul.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.mag.acu <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.mag.acu,
family = "bernoulli",
site.weights = mag.acu.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.01,
bag.fraction = 0.5)

myGbmModelOut.ost.vir <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.ost.vir,
family = "bernoulli",
site.weights = ost.vir.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.001,
bag.fraction = 0.5)

myGbmModelOut.pin.str <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.pin.str,
family = "bernoulli",
site.weights = pin.str.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.01,
bag.fraction = 0.5)

myGbmModelOut.pla.occ <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.pla.occ,
family = "bernoulli",
site.weights = pla.occ.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.001,
bag.fraction = 0.5)

myGbmModelOut.pru.ser <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.pru.ser,
family = "bernoulli",
site.weights = pru.ser.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0025,
bag.fraction = 0.5)

myGbmModelOut.que.alb <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.que.alb,
family = "bernoulli",
site.weights = que.alb.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.005,
bag.fraction = 0.5)

myGbmModelOut.que.vel <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.que.vel,
family = "bernoulli",
site.weights = que.vel.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.005,
bag.fraction = 0.5)

myGbmModelOut.til.ame <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.til.ame,
family = "bernoulli",
site.weights = til.ame.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.02,
bag.fraction = 0.5)

myGbmModelOut.tsu.can <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.tsu.can,
family = "bernoulli",
site.weights = tsu.can.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.02,
bag.fraction = 0.5)

myGbmModelOut.ulm.ame <- gbm.step(
data=myData,
gbm.x = myExpl,
gbm.y = myRespName.ulm.ame,
family = "bernoulli",
site.weights = ulm.ame.weights,
tree.complexity = myTreeComplexity,
learning.rate = 0.0075,
bag.fraction = 0.5)



#Save workspace.

save.image(file="sdm.objects.with.Native.variables.biomod2.step.1.RData")