nordpred.estimate: Fit power5 and poisson Age-Period-Cohort models for...
In haraldwf/nordpred: Prediction of cancer incidence (as used in the Nordpred project)

Description Usage Arguments Details Value Note for S-plus Author(s) References See Also Examples

nordpred.estimate estimates parameters in the power5 or poisson Age-Period-Cohort (APC) model

nordpred.estimate(cases, pyr, noperiod, startestage, linkfunc = "power5",
  find_best_model = FALSE)

is.nordpred.estimate(x)

`cases`	A `data.frame` with number of cases
`pyr`	A `data.frame` with observed and forecasted person years.
`noperiod`	The number of periods to be used in prediction base.
`startestage`	Youngest age group to include in the regression model
`linkfunc`	Link function to use in the model. Default is special version used in the Nordpred project ('power5'), where the link is g(x) = x^0.2, while the alternative is the poisson function ('poisson'), where the link is g(x) = log(x)
`find_best_model`	Logical
`x`	an object to test for class `nordpred`

For details of the choice of prediction base, significance test for using recent slope, and for the power5 model, see Moller B., Fekjaer H. et al. (2002), see references.

nordpred.estimate returns an object of class nordpred.estimate (see nordpred.estimate.object).

is.nordpred.estimate returns TRUE if input object is of class nordpred.estimate, FALSE otherwise.

Powerlink is made via a special modification in S-PLUS. This works fine for the point estimates, but the variance estimates found via the glm-objects are wrong. For variance estimates, we would rather recommend using R.

Harald Fekjaer and Bjorn Moller (Cancer Registry of Norway)

A website for nordpred is available at: http://www.kreftregisteret.no/software/nordpred/
Background for the methods can be found in: Moller B., Fekjaer H., Hakulinen T., Sigvaldason H, Storm H. H., Talback M. and Haldorsen T 'Prediction of cancer incidence in the Nordic countries: Empirical comparison of different approaches' Statistics in Medicine 2003; 22:2751-2766
An application of the function, using all the default settings, can be found in: Moller B, Fekjaer H, Hakulinen T, Tryggvadottir L, Storm HH, Talback M, Haldorsen T. Prediction of cancer incidence in the Nordic countries up to the year 2020. Eur J Cancer Prev Suppl 2002; 11: S1-S96

Other nordpred: is.nordpred, nordpred; nordpred.estimate.object; nordpred.getpred; nordpred.object; nordpred.prediction; plot.nordpred; print.nordpred.estimate; print.nordpred; summary.nordpred

# data (Colon cancer for Norwegian males)
indata

# Create dataset with observed and predicted population
inpop <- cbind(inpop1, inpop2)

# Fit model using powerlink (default):
est <- nordpred.estimate(cases = indata, pyr = inpop, noperiod = 4, startestage = 5)

# Fit model using poisson link:
est2 <- nordpred.estimate(indata, inpop, 4, 5, linkfunc = 'poisson')

# Print results:
print(est)
print(est$glm)

#' # Use estimat object to make predictions:
res <- nordpred.prediction(est, startuseage = 6, cuttrend = c(0, .25, .5, .75, .75), recent = TRUE)