Kaggle Competition Walkthrough: Fitting a model

1. Parameters.R

####################################
# Training parameters
####################################
MyTrainControl=trainControl(
		method = "repeatedCV",
		number=10,
		repeats=5,
		returnResamp = "all",
		classProbs = TRUE,
		summaryFunction=twoClassSummary
		)

2. fit model.R

model <- train(FL,data=trainset,method='glmnet',
	metric = "ROC",
	tuneGrid = expand.grid(.alpha=c(0,1),.lambda=seq(0,0.05,by=0.01)),
	trControl=MyTrainControl)
model
plot(model, metric='ROC')

3. model output.txt

> model
250 samples
200 predictors
  2 classes: 'X0', 'X1' 

Pre-processing: None 
Resampling: Cross-Validation (10 fold, repeated 1 times) 

Summary of sample sizes: 225, 226, 225, 225, 225, 225, ... 

Resampling results across tuning parameters:

  alpha  lambda  Sens   Spec   ROC    Sens SD  Spec SD  ROC SD
  0      0       0.731  0.802  0.827  0.102    0.121    0.0888
  0      0.01    0.731  0.802  0.827  0.102    0.121    0.0888
  0      0.02    0.764  0.741  0.829  0.117    0.154    0.0863
  0      0.03    0.698  0.817  0.827  0.131    0.109    0.0819
  0      0.04    0.764  0.718  0.825  0.124    0.161    0.0825
  0      0.05    0.681  0.825  0.826  0.155    0.13     0.0793
  1      0       0.722  0.688  0.792  0.126    0.201    0.0947
  1      0.01    0.673  0.749  0.756  0.112    0.112    0.0691
  1      0.02    0.798  0.527  0.729  0.105    0.194    0.0663
  1      0.03    0.539  0.748  0.69   0.156    0.189    0.0648
  1      0.04    0.84   0.382  0.681  0.114    0.136    0.0616
  1      0.05    0.48   0.746  0.662  0.243    0.235    0.0627

ROC was used to select the optimal model using  the largest value.
The final values used for the model were alpha = 0 and lambda = 0.02. 

4. Evaluate model.R

test <- predict(model, newdata=testset, type  = "prob")
colAUC(test, testset$Target)

5. Multicore.R

####################################
# Setup Multicore
####################################
#source:
#http://www.r-bloggers.com/feature-selection-using-the-caret-package/
if ( require("multicore", quietly = TRUE, warn.conflicts = FALSE) ) {
	MyTrainControl$workers <- multicore:::detectCores()
	MyTrainControl$computeFunction <- mclapply
	MyTrainControl$computeArgs <- list(mc.preschedule = FALSE, mc.set.seed = FALSE)
}