How to make predictions using XGBoost - r

Question

I am trying to make my first model using XGBoost and I cannot figure out how to actually get my prediction values. I was able to train a model and get root means squared error values, but I don't know where to go from here.
My dataset is about house prices. I am using variables such as: LotFrontage, LotArea, BldgType, OverallQual, OverallCond, FullBath, HalfBath, TotRmsAbvGrd, YearBuilt, TotalBsmtSF, BedroomAbvGr, and GrLivArea. Some of these variables are numeric and some are strings.
Here is my code and where I am getting an error:
library(data.table)
library(caret)
library(Metrics)
library(xgboost)
train<-fread("train_data.csv")
test<-fread("test_data.csv")
sub_train<-train[,.(LotFrontage,LotArea,BldgType,OverallQual,OverallCond,FullBath,HalfBath,TotRmsAbvGrd,YearBuilt,TotalBsmtSF,BedroomAbvGr,GrLivArea,SalePrice)]
sub_test<-test[,.(LotFrontage,LotArea,BldgType,OverallQual,OverallCond,FullBath,HalfBath,TotRmsAbvGrd,YearBuilt,TotalBsmtSF,BedroomAbvGr,GrLivArea)]
sub_test$SalePrice<-0
y.train<-sub_train$SalePrice
y.test<-sub_test$SalePrice
dummies <- dummyVars(SalePrice~ ., data = sub_train)
x.train<-predict(dummies, newdata = sub_train)
x.test<-predict(dummies, newdata = sub_test)
dtrain <- xgb.DMatrix(x.train,label=y.train,missing=NA)
dtest <- xgb.DMatrix(x.test,label=y.test,missing=NA)
param <- list( objective = "reg:linear",
gamma =0.02,
booster = "gbtree",
eval_metric = "rmse",
eta = 0.02,
max_depth = 10,
subsample = 0.9,
colsample_bytree = 0.9,
tree_method = 'hist'
)
XGBm<-xgb.cv( params=param,nfold=5,nrounds=2000,missing=NA,data=dtrain,print_every_n=1)
pred<-predict(XGBm, sub_test$SalePrice)
watchlist <- list(eval = dtest, train = dtrain)
XGBm<-xgb.train( params=param,nrounds=200,missing=NA,data=dtrain,watchlist,early_stop_round=20,print_every_n=1)
sub_train2 <- xgb.DMatrix(x.train,label=y.train,missing=NA)
pred1<-predict(XGBm, sub_train$SalePrice)
Here is a screenshot of my error:
So, I would like to get a csv file full of predicted house prices. I want to update the SalePrice column within the train dataset or the sub_train dataset like sub_train$SalePrice<-predict(XGBoost,sub_train$SalePrice). Any ideas?
Also, I have gotten a "predict" line to run, but it just gives me decimals like .823 and .174 and so on, and that is not what I am looking for. I want house prices with values over 100,000.
Thanks!