DIAGNOSTICS {nsRFA}R Documentation

Diagnostics of models

Description

Diagnostics of model results, it compares estimated values y with observed values x.

Usage

 R2 (x, y, na.rm=FALSE)
 RMSE (x, y, na.rm=FALSE) 
 MAE (x, y, na.rm=FALSE)
 RMSEP (x, y, na.rm=FALSE)
 MAEP (x, y, na.rm=FALSE)

Arguments

x

observed values

y

estimated values

na.rm

logical. Should missing values be removed?

Details

If xi are the observed values, yi the estimated values, with i=1,...,n, and <x> the sample mean of xi, then:

R2 = 1 - (sum(xi-yi)^2)/(sum(xi^2-n<x>^2)

RMSE = sqrt(1/n sum(xi-yi)^2)

MAE = 1/n sum(|xi-yi|)

RMSEP = sqrt(1/n sum((xi-yi)/xi)^2)

MAEP = 1/n sum(|(xi-yi)/xi|

See http://en.wikipedia.org/wiki/Coefficient_of_determination, http://en.wikipedia.org/wiki/Mean_squared_error and http://en.wikipedia.org/wiki/Mean_absolute_error for other details.

Value

R2 returns the coefficient of determination R2 of a model.

RMSE returns the root mean squared error of a model.

MAE returns the mean absolute error of a model.

RMSE returns the percentual root mean squared error of a model.

MAE returns the percentual mean absolute error of a model.

Note

For information on the package and the Author, and for all the references, see nsRFA.

See Also

lm, summary.lm, predict.lm, REGRDIAGNOSTICS

Examples

data(hydroSIMN)

datregr <- parameters
regr0 <- lm(Dm ~ .,datregr); summary(regr0)
regr1 <- lm(Dm ~ Am + Hm + Ybar,datregr); summary(regr1)

obs <- parameters[,"Dm"]
est0 <- regr0$fitted.values
est1 <- regr1$fitted.values

R2(obs, est0)
R2(obs, est1)

RMSE(obs, est0)
RMSE(obs, est1)

MAE(obs, est0)
MAE(obs, est1)

RMSEP(obs, est0)
RMSEP(obs, est1)

MAEP(obs, est0)
MAEP(obs, est1)

[Package nsRFA version 0.7-14 Index]