polyreg (version 0.6.3)

polyFit: Polynomial Fit

Description

Fit polynomial regression using a linear or logistic model; predict new data.

Usage

polyFit(xy, deg, maxInteractDeg=deg, use = "lm", pcaMethod=NULL,
     pcaLocation='front', pcaPortion=0.9, glmMethod="one", 
     return_xy=FALSE, returnPoly=FALSE)
# S3 method for polyFit
predict(object, newdata, ...)

Arguments

xy

Data frame with response variable in the last column. In the classification case, response is class ID, stored in a vector, not as a factor. Categorical variables (> 2 levels) should be passed as factors, not dummy variables or integers, to ensure the polynomial matrix is constructed properly.

deg

The max degree for polynomial terms.

maxInteractDeg

The max degree of interaction terms.

use

Set to 'lm' for linear regression, 'glm' for logistic regression, or 'mvrlm' for multivariate-response lm.

pcaMethod

NULL for no PCA. For PCA, can be either 'prcomp' (use the prcomp function) or 'RSpectra' (use the eigs function in the RSpectra package.

pcaLocation

In case PCA is applied, specify 'front' to have PCA calculated before forming polynomials, otherwise 'back.

pcaPortion

If less than 1.0, use as many principal components so as to achieve this portion of total variance. Otherwise, use this many components. In the 'RSpectra' case, this value must be an integer of 1 or more.

glmMethod

Defaults to "one."

newdata

Data frame, one row for each "X" to be predicted. Must have the same column names as in xy (without "Y").

object

An item of class 'polyFit' containing output. Can be used with predict().

return_xy

return data? Default: FALSE

returnPoly

return polyMatrix object? Defaults to FALSE since may be quite large.

...

Additional arguments for getPoly().

Value

The return value of polyFit() is an polyFit object. The orginal arguments are retained, along with the fitted models and so on.

The prediction function predict.polyFit returns the predicted value(s) for newdata. In the classification case, these will be the predicted class labels, 1,2,3,...

Details

The polyFit function calls getPoly to generate polynomial terms from predictor variables, then fits the generated data to a linear or logistic regression model. (Powers of dummy variables will not be generated, other than degree 1, but interaction terms will calculated.

If pcaMethod is not NULL, a principal component analysis is performed before or after generating the polynomials.

When logistic regression for classification is indicated, with more than two classes, All-vs-All or One-vs-All methods, coded 'all' and 'one', can be applied to deal with multiclass problem. Multinomial logit ('multilog') is also available.

Under the 'mvrlm' option in a classification problem, lm is called with multivariate response, using cbind and dummy variables for class membership as the response. Since predictors are used to form polynomials, this should be a reasonable model, and is much faster than 'glm'.

Examples

# NOT RUN {
N <- 125
xyTrain <- data.frame(x1 = rnorm(N), 
                      x2 = rnorm(N),
                      group = sample(letters[1:5], N, replace=TRUE),
                      score = sample(100, N, replace = TRUE) # final column is y
                      )

pfOut <- polyFit(xyTrain, 2)

# 4 new test points
xTest <- data.frame(x1 = rnorm(4), 
                    x2 = rnorm(4),
                    group = sample(letters[1:5], 4, replace=TRUE))
  
predict(pfOut, xTest) # returns vector of 4 predictions


# spot checks
stopifnot(length(predict(pfOut, xTest)) == nrow(xTest))

data(prgeng) # US Census engineer wage data 
indx <- seq(1, nrow(prgeng), 10)
xy <- prgeng[indx, c(1:4,6,5)] # Y value (wageinc here) must be last
xy <- toFactors(xy, 2:4)  # convert education, occupation, and sex to factors
                          # (don't pass categorical variables as integer codes when fitting)

# regression example
pfout <- polyFit(xy, 2)  # try degree 2 model

# say we want to predict wage for a woman, age 35, some college,
# occupation 101, worked 52 weeks
newx <- data.frame(age=35, educ=11, occ=101, sex=2, wkswrkd=52) 
# predict(pfout,newx)  # 38830.54 

# classification example, logistic
xy <- cbind(xy[,-3], xy$occ)  # Y value (occupaton here) must be last
pfout <- polyFit(xy,2,use='glm')  # try degree 2 model
# say we want to predict occupation for a woman, age 35, college grad,
# occupation 101, worked 52 weeks; and for a man with the same
# characteristics
newx <- data.frame(age=c(35,35),
                   educ=c(13,13),
                   sex=c(2,1),
                   wageinc=c(62000,62000),
                   wkswrkd=c(52,52)) 
preds <- predict(pfout,newx)  
levels(xy$occ)[preds]  # translate from class number; 102, 102



# }