hmd.DA.mx {LifeTables}

Model Life Table Discriminant Analysis

Description

Training and testing steps to perform a discriminant analysis for a partial or complete human mortality rate schedule with the Human Mortality Database as the training set.

Usage

hmd.DA.mx(data, age, sex="female")

Arguments

data

Partial or complete human mortality rate schedule(s) to be classified. The set of age-specific mortality rates to be classified must match up with one of the age groups from the HMD data (standard 5-year age intervals 0, 1-4, 5-9, 10-14...110+) and be in the original scale (non-logged), i.e. one could classify mortality rates for ages 0, 1-5, 10-14, 20-24, 80-84, but cannot classify mortality rates for ages 0-5, 20-29, 30-39, etc. This argument can be either a single schedule or multiple schedules, but either way data must be a matrix with ages in columns.

age

The ages of the mortality rates to be classified. This argument tells the function which mortality rates from the HMD should be used for the training models. age is a vector of length equal to the number of columns in data with the starting ages of the intervals to be classified. (e.g. if the user wants to classify mortality rates from ages 20-24, 25-29, 30-34, 35-39, 50-54, enter age=c(seq(20,35,5),50)).

sex

"male" or "female". The sex for the schedule to be classified.

Value

train

The output from the function MclustDA, the parameters and other summary information for the model best fitting each class according to BIC.

out.dens

The output from the function predict(MclustDA model name), A matrix in which the [i,j]th entry is the density for observation i in the model for class j.

classification

The classification for the test data

Author(s)

David J. Sharrow

References

S. Clark and D. Sharrow (2011) "Contemporary Model Life Tables for Developed Countries: An Application of Model-Based Clustering." Paper presented at the annual meeting of the Population Association of America. http://paa2011.princeton.edu/abstracts/111808

C. Fraley and A. E. Raftery (2002) "Model-based clustering, discriminant analysis, and density estimation." Journal of the American Statistical Association 97:611-631.

C. Fraley, A. E. Raftery, T. B. Murphy, L. Scrucca (2012) "mclust Version 4 for R: Normal Mixture Modeling for Model-Based Clustering, Classification, and Density Estimation", Technical Report no. 597, Department of Statistics, University of Washington

Human Mortality Database University of California, Berkeley (USA), and Max Planck Institute for Demographic Research (Germany). Available at www.mortality.org or www.humanmortality.de (data downloaded February 2015).

See Also

MclustDA, Mclust, MLTobs

Examples

# some test data
data(MLTobs)

##48 Belgium 1860-64 (known class = 1)
##180 England 1925-29 (known class = 2)
##207 Estonia 2005-09 (known class = 7)
##266 France 1960-64 (known class = 3)
##410 Japan 2000-04 (known class = 5)
##607 Russia 1980-84 (known class = 6)
##798 USA 2000-04 (known class = 4)

country.nums <- c(48,180,207,266,410,607,798)
test.mx <- t(flt.mx[3:10,country.nums]) # mortality rates for ages 5-40
test.age <- seq(5,40,5)

# classify the test data matrix
examp.out <- hmd.DA.mx(data=test.mx, age=test.age, sex="female")
examp.out$classification

# classify the test data single schedule as matrix
examp.out2 <- hmd.DA.mx(data=t(as.matrix(test.mx[4,])), age=test.age, sex="female")
examp.out2$classification