PredefinedModels: Classification and Regression Models
In juscelino-izidoro/supcavs: Supervised and Unsupervised PCA Variable Selection

Description Usage Arguments Details Value Examples

linearRegressionModel build a multiple linear regression and returns r-squared. artificialNeuralNetworkModel and naiveBayesModel do classification and return iterative test error and hit ratio, respectively.

1
2
3

linearRegressionModel(x, y, pcaMethod="prcomp", pcaParams=NULL)
artificialNeuralNetworkModel(x, y, pcaMethod=NULL, pcaParams=NULL)
naiveBayesModel(x, y, pcaMethod=NULL, pcaParams=NULL)

`x`	a data frame with explanatory variables obtained from a data set.
`y`	a vector with response vector
`pcaMethod`	a string that specify the name of pca function. By default, this is defined as "prcomp"
`pcaParams`	a vector with param=value elements for pcaMethod.

linearRegressionModel executes a multiple linear regression in a way that "y" is regressed by variables in data frame "x".

artificialNeuralNetworkModel executes a binary pattern recognition for classification task. It uses 66% of the data to train an ANN and uses 34% to test the model. On the training, this function uses the variables in "x" (the columns of data.frame) to predict the response variable "y".

naiveBayesModel executes a binary pattern classification. It uses 66% of the data to train a NB and uses 34% to test the model. On the training, this functions uses the variables in "x" as predictors and "y" as the response variable.

Each function returns just one value:

linearRegressionModel returns r-squared coefficient.

artificialNeuralNetworkModel returns iterative test error.

naiveBayesModel returns hit ratio.

x = as.data.frame(matrix(rnorm(1000),
                  100, 10, 
                  dimnames=list(NULL, paste('X', 1:10, sep=''))))
                  
y = rowSums(x) + rnorm(100)

# $R^2$
linearRegressionModel(x,y)

# Sum of Squared Error
library('RSNNS')
artificialNeuralNetworkModel(x,y)

# Hit ratio
library('e1071')
naiveBayesModel(x,y)