rxLinPredError: Calculates prediction error statistics for linear regression...

In RevoEnhancements/RevoEnhancements: Selected enhancements to Revolution Analytics RevoScaleR

Description

Calculates a number of error statistics, including:

MSE

Mean squared error

MAPE

Mean absolute percentage erorr

MPE

Mean percentage error

MSWD

Mean squared weighted deviation

Usage

rxLinPredError(actualVarName, predVarName, data, sWeights = NULL,
  blocksPerRead = 1, reportProgress = rxGetOption("reportProgress"))

Arguments

actualVarName	String name of the response variable.
predVarName	String name of the predicted variable.
sWeights	String name of error weights.
data	data frame, or character string containing an '.xdf' file name (with path), or RxXdfData object representing an '.xdf' file containing the actual and observed variables.
blocksPerRead	number of blocks to read for each chunk of data read from the data source.
reportProgress	Passed to rxDataStep

Value

returns a list of prediction measures MSE, MAPE, MPE, MSWD

See Also

Other Model summary statistics: rxAIC; rxF1score

Examples

library(RevoScaleR)

## Demonstrates calculation on data frame

fit <- rxLinMod(Sepal.Length ~ Petal.Length + Petal.Width, data = iris)
prd <- rxPredict(fit, iris)$Sepal.Length_Pred
dat <- data.frame(Sepal.Length=iris$Sepal.Length, Sepal.Length_Pred=prd, Weights = rep(1, nrow(iris))/nrow(iris))
rxLinPredError("Sepal.Length", "Sepal.Length_Pred", data=dat, sWeights="Weights")
rxLinPredError("Sepal.Length", "Sepal.Length_Pred", data=dat)

RevoEnhancements/RevoEnhancements documentation built on May 9, 2019, 9:46 a.m.