News for Package AppliedPredictiveModeling

Changes in version 1.1-6

• The file CreateGrantData.R was updated to include code to create the objects factorPredictors and factorForm.

Changes in version 1.1-5

The package dependencies were updated. Some were moved to 'Imports'

The chapter scripts were re-run with the latest versions of the AppliedPredictiveModeling andcaret packages. For caret, the names of the tuning parameter columns were changed to remove the dot. These were made to be consistent with the newer version of caret and are not required. For example, .sigma was changed to sigma and so on.

Additional changes are:

• 03_Data_Pre_Processing.R was changed to include code from the Computing section on creating dummy variables.

• 04_Over_Fitting.R was modified. Changes were made to:

  • avoid a warning message when the SVM grid was created

  • classProbs = TRUE was added to the control function for the object svmFit

  • some notes were made in the potential differences in SVM results between versions of caret

• 06_Linear_Regression.R was updated to use the newer "ridge" model in caret.

• In 07_Non-Linear_Reg.R, some notes were made in the potential differences in SVM results between versions of caret

• In 16_Class_Imbalance.R, verboseness (verbosity?) was turned off to make the results cleaner. Also, the code for svmWtFit was using the wrong tuning grid (svmGrid2 instead of svmGrid1).

• 19_Feature_Select.R was changed so that the resampling values were only saved for the final model.

Changes in version 1.1-4

• The data set ChemicalManufacturingProcess did not contain the rows with missing data. They were added back in.

• Small changes to conform to R CMD check.

Changes in version 1.1-2

• Code to create the carsSubset object in Seciton 3.8 was added to 03_Data_Pre_Processing.R

Changes in version 1.1-1

• Initial Version