makeTower {toweranNA}

Nonimputational method for dealing with NA values in prediction application

Description

In a prediction application, the intended regression model is applied to complete cases, from which marginal regression models can be derived for predicting new cases having arbitrary NA patterns.

Usage

makeTower(data,yName,regFtnName,opts,scaling=NULL,yesYVal=NULL)
## S3 method for class 'tower'
predict(object,newx,k=1,...)

Arguments

data

Data frame or equivalent.

yName

Name of the column in data to be predicted.

regFtnName

Regression model to be used, currently 'lm', 'glm' (family=binomial), or 'towerKNN'.

opts

Optional arguments for regFtnName, an R list.

k

number of nearest neighbors

scaling

Scaling to be applied to x and newx. Default NULL means no scaling.

yesYVal

In the case of dichotomous Y, this specifies the level to be considered positive, i.e. for which Y will be 1.

object

Object of type 'tower'.

newx

New case(s) to be predicted, in the same format as in the non-Y portion of data.

...

Other arguments need by regFtnName.

Value

Object of class 'tower', to be used as input to predict.tower.

Author(s)

Norm Matloff, Pete Mohanty

Examples

towerOut <- makeTower(mtcars,'mpg','lm')

newx <- mtcars[-c(1:10),-1]
for(i in 1:10)
  newx[i, i] <- NA

head(newx)
#                     cyl  disp  hp drat    wt  qsec vs am gear carb
# Merc 280C            NA 167.6 123 3.92 3.440 18.90  1  0    4    4
# Merc 450SE            8    NA 180 3.07 4.070 17.40  0  0    3    3
# Merc 450SL            8 275.8  NA 3.07 3.730 17.60  0  0    3    3
# Merc 450SLC           8 275.8 180   NA 3.780 18.00  0  0    3    3
# Cadillac Fleetwood    8 472.0 205 2.93    NA 17.98  0  0    3    4
# Lincoln Continental   8 460.0 215 3.00 5.424    NA  0  0    3    4
 
predict(towerOut,newx,k=3)
#  [1] 20.00086 15.17132 15.17132 15.17132 11.15469 11.15469 11.15469 28.52625
#  [9] 29.06067 28.52625 24.72144 17.45622 16.75827 15.52077 14.95958 28.52625
# [17] 25.34890 26.08506 15.52077 19.19484 15.37239 24.72144