nobs: Extract the number of observations to which a VLMC model is...

In PST: Probabilistic Suffix Trees and Variable Length Markov Chains

Extract the number of observations to which a VLMC model is fitted

Description

The number of observations to which a VLMC model is fitted is notably used for computing the Bayesian information criterion BIC or the Akaike information criterion with correction for finite sample sizes AICc.

Usage

## S4 method for signature 'PSTf'
nobs(object)

Arguments

object

A PST, that is an object of class PSTf as returned by the pstree or prune method.

Details

This is the method for the generic nobs function provided by the stats4 package. The number of observations to which a VLMC model is fitted is the total number of symbols in the learning sample. If the learning sample contains missing values and the model is learned without including missing values (see pstree), the total number of symbols is the number of non-missing states in the sequence(s). This information is used to compute the Bayesian information criterion of a fitted VLMC model. The BIC generic function calls the logLik and nobs methods for class PSTf. For more details, see Gabadinho 2016.

Value

An integer containing the number of symbols in the learning sample.

Author(s)

Alexis Gabadinho

References

Gabadinho, A. & Ritschard, G. (2016). Analyzing State Sequences with Probabilistic Suffix Trees: The PST R Package. Journal of Statistical Software, 72(3), pp. 1-39.

See Also

BIC

Examples

data(s1)
s1.seq <- seqdef(s1)
S1 <- pstree(s1.seq, L=3)
nobs(S1)

## Self rated health sequences
## Loading the 'SRH' data frame and 'SRH.seq' sequence object
data(SRH)

## model without considering missing states
## model with max. order 2 to reduce computing time
## nobs is the same whatever L and nmin
m1 <- pstree(SRH.seq, L=2, nmin=30, ymin=0.001)
nobs(m1)

## considering missing states, hence nobs is higher
m2 <- pstree(SRH.seq, L=2, nmin=30, ymin=0.001, with.missing=TRUE)
nobs(m2)

PST documentation built on June 22, 2024, 6:50 p.m.