jackStats: Jackknife Statistics
In USGS-R/rloadest: River Load Estimation

Description Usage Arguments Value Note References See Also Examples

Compute selected jackknife statistics for a rating-curve load-estimation model.

1	jackStats(fit, which = "load")

`fit`	an object of class "loadReg"—output from `loadReg`. Can also be an object of class "censReg."
`which`	a character string indicating the "load" or "concentration" model for an object of class "loadReg" or "censReg" for an object of class "censReg."

An object of class "jackStats" containing these components: coef, the table of coefficient estimates, the jackknife bias and standard errors
coefficients, the jackknifed coefficients
pctcens, the percentage of left-censored values.
The PRESS statistic and individual jackknife differences are also returned when the percentage of censoring is 0.

The jackStats function can only be used when the analysis is AMLE.

Abdi and Williams (2010) describe the jackknife as refering to two related techniques: the first estimates the parameters, their bias and standard errors and the second evaluates the predictive performance of the model. The second technique is the PRESS statistic (Helsel and Hirsch, 2002), but can only be used on uncensored data; it is computed by jackStats when no data are censored. The first technique can be used to assess the coefficients of the regression—the bias should be small and the jackknife standard errors should not be much different from the standard errors reported for the regression. Efron and Tibshirani (1993) suggest that the bias is small if the relative bias (biuas divided by the jackknife standard error) is less than 0.25.

Abdi, H. and Williams, L.J., 2010, Jackknife, in encyclopedia of research design, Salkind, N.J., editor: Thousand Oaks, Calif., SAGE Publications, 1719 p.

Efron, B. and Tibshirani, R.J., 1993, An introduction to the bootstrap: Boca Raton, Fla., Chapman and Hall/CRC, 436 p.

Helsel, D.R. and Hirsch, R.M., 2002, Statistical methods in water resources: U.S. Geological Survey Techniques of Water-Resources Investigations, book 4, chap. A3, 522 p. Salkind,

loadReg

# From application 1 in the vignettes
data(app1.calib)
app1.lr <- loadReg(Phosphorus ~ model(1), data = app1.calib, 
 flow = "FLOW", dates = "DATES", conc.units="mg/L",
 station="Illinois River at Marseilles, Ill.")
jackStats(app1.lr)