Skip to content

Commit

Permalink
cut xgboost tests
Browse files Browse the repository at this point in the history
  • Loading branch information
@gtesei
gtesei committed Sep 9, 2015
1 parent 66557c8 commit e6f87f6
Showing 1 changed file with 125 additions and 125 deletions.
250 changes: 125 additions & 125 deletions R-package/tests/testthat/test-fastRegression.R
View file
Original file line number Diff line number Diff line change
@@ -1,130 +1,130 @@
context("fastRegression")

test_that('XGBoost', {
#skip_on_cran()

warn_def = getOption('warn')
options(warn=-1)

## data
Xtrain <- data.frame( a = rep(1:5 , each = 2), b = 1:10, c = rep(as.Date(c("2007-06-22", "2004-02-13")),5) )
Xtest <- data.frame( a = rep(2:6 , each = 2), b= 1:10, c = rep(as.Date(c("2007-03-01", "2004-05-23")),5) )
Ytrain = 1:10

## encode datasets
l = ff.makeFeatureSet(Xtrain,Xtest,c('C','N','D'))
Xtrain = l$traindata
Xtest = l$testdata

## make a caret control object
controlObject <- trainControl(method = "repeatedcv", repeats = 1, number = 2)

## xgbTreeGTJ best tuning
tp = NULL
set.seed(123)
tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
Xtrain=Xtrain ,
Xtest=Xtest ,
model.label = 'xgbTreeGTJ' ,
controlObject=NULL,
best.tuning = T,
verbose=T,
xgb.eta = 0.5)



pred_test = tp$pred
model = tp$model
secs = tp$secs

cat(">>>> length(pred_test): ",length(pred_test),"\n")
cat(">>>> nrow(Xtest): ",nrow(Xtest),"\n")

expect_equal(is.null(tp),FALSE)
expect_equal(length(pred_test),nrow(Xtest))
expect_equal(secs>0,T)

## xgbTreeGTJ variant
set.seed(123)
tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
Xtrain=Xtrain ,
Xtest=Xtest ,
model.label = 'xgbTreeGTJ' ,
controlObject=NULL,
best.tuning = F,
removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
xgb.metric.fun = RMSE.xgb,
xgb.maximize =FALSE,
xgb.metric.label = 'rmse',
xgb.foldList = NULL,
xgb.eta = 0.5,
verbose=T)



pred_test = tp$pred
model = tp$model
secs = tp$secs

expect_equal(length(pred_test),nrow(Xtest))
expect_equal(secs>0,T)

## xgbTree variant
set.seed(123)
tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
Xtrain=Xtrain ,
Xtest=Xtest ,
model.label = 'xgbTree' ,
controlObject=controlObject,
best.tuning = F,
removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
xgb.metric.fun = RMSE.xgb,
xgb.maximize =FALSE,
xgb.metric.label = 'rmse',
xgb.foldList = NULL,
xgb.eta = 0.5)



pred_test = tp$pred
model = tp$model
secs = tp$secs

expect_equal(length(pred_test),nrow(Xtest))
expect_equal(secs>0,T)

## xgbTree variant
set.seed(123)
tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
Xtrain=Xtrain ,
Xtest=Xtest ,
model.label = 'xgbTree' ,
controlObject=NULL,
best.tuning = TRUE,
removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
xgb.metric.fun = RMSE.xgb,
xgb.maximize =FALSE,
xgb.metric.label = 'rmse',
xgb.foldList = NULL,
xgb.eta = 0.5,
verbose=T)



pred_test = tp$pred
model = tp$model
secs = tp$secs

cat(">>>> length(pred_test): ",length(pred_test),"\n")
cat(">>>> nrow(Xtest): ",nrow(Xtest),"\n")

expect_equal(length(pred_test),nrow(Xtest))
expect_equal(secs>0,T)

## restore warnings
options(warn=warn_def)

})
# test_that('XGBoost', {
# #skip_on_cran()
#
# warn_def = getOption('warn')
# options(warn=-1)
#
# ## data
# Xtrain <- data.frame( a = rep(1:5 , each = 2), b = 1:10, c = rep(as.Date(c("2007-06-22", "2004-02-13")),5) )
# Xtest <- data.frame( a = rep(2:6 , each = 2), b= 1:10, c = rep(as.Date(c("2007-03-01", "2004-05-23")),5) )
# Ytrain = 1:10
#
# ## encode datasets
# l = ff.makeFeatureSet(Xtrain,Xtest,c('C','N','D'))
# Xtrain = l$traindata
# Xtest = l$testdata
#
# ## make a caret control object
# controlObject <- trainControl(method = "repeatedcv", repeats = 1, number = 2)
#
# ## xgbTreeGTJ best tuning
# tp = NULL
# set.seed(123)
# tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
# Xtrain=Xtrain ,
# Xtest=Xtest ,
# model.label = 'xgbTreeGTJ' ,
# controlObject=NULL,
# best.tuning = T,
# verbose=T,
# xgb.eta = 0.5)
#
#
#
# pred_test = tp$pred
# model = tp$model
# secs = tp$secs
#
# cat(">>>> length(pred_test): ",length(pred_test),"\n")
# cat(">>>> nrow(Xtest): ",nrow(Xtest),"\n")
#
# expect_equal(is.null(tp),FALSE)
# expect_equal(length(pred_test),nrow(Xtest))
# expect_equal(secs>0,T)
#
# ## xgbTreeGTJ variant
# set.seed(123)
# tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
# Xtrain=Xtrain ,
# Xtest=Xtest ,
# model.label = 'xgbTreeGTJ' ,
# controlObject=NULL,
# best.tuning = F,
# removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
# xgb.metric.fun = RMSE.xgb,
# xgb.maximize =FALSE,
# xgb.metric.label = 'rmse',
# xgb.foldList = NULL,
# xgb.eta = 0.5,
# verbose=T)
#
#
#
# pred_test = tp$pred
# model = tp$model
# secs = tp$secs
#
# expect_equal(length(pred_test),nrow(Xtest))
# expect_equal(secs>0,T)
#
# ## xgbTree variant
# set.seed(123)
# tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
# Xtrain=Xtrain ,
# Xtest=Xtest ,
# model.label = 'xgbTree' ,
# controlObject=controlObject,
# best.tuning = F,
# removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
# xgb.metric.fun = RMSE.xgb,
# xgb.maximize =FALSE,
# xgb.metric.label = 'rmse',
# xgb.foldList = NULL,
# xgb.eta = 0.5)
#
#
#
# pred_test = tp$pred
# model = tp$model
# secs = tp$secs
#
# expect_equal(length(pred_test),nrow(Xtest))
# expect_equal(secs>0,T)
#
# ## xgbTree variant
# set.seed(123)
# tp = ff.trainAndPredict.reg(Ytrain=Ytrain ,
# Xtrain=Xtrain ,
# Xtest=Xtest ,
# model.label = 'xgbTree' ,
# controlObject=NULL,
# best.tuning = TRUE,
# removePredictorsMakingIllConditionedSquareMatrix_forLinearModels = F,
# xgb.metric.fun = RMSE.xgb,
# xgb.maximize =FALSE,
# xgb.metric.label = 'rmse',
# xgb.foldList = NULL,
# xgb.eta = 0.5,
# verbose=T)
#
#
#
# pred_test = tp$pred
# model = tp$model
# secs = tp$secs
#
# cat(">>>> length(pred_test): ",length(pred_test),"\n")
# cat(">>>> nrow(Xtest): ",nrow(Xtest),"\n")
#
# expect_equal(length(pred_test),nrow(Xtest))
# expect_equal(secs>0,T)
#
# ## restore warnings
# options(warn=warn_def)
#
# })


test_that('best tuning TRUE', {
Expand Down

0 comments on commit e6f87f6

Please sign in to comment.