Clustering algorithm
\ ˈklʌstərɪŋ \ ˈælgəˌrɪðəm \
A technique used to cluster together values of a rating factor into groups exhibiting similar characteristics , so that the differences between groups become statistically significant.
When using rating factors to model longevity, sometimes a factor naturally takes more values than can be modelled effectively with the given volume of data available. In such circumstances it is necessary to cluster some values of the rating factor together to increase the amount of data available in each group and increase the statistical significance of the resulting model.
For example, if using address as a rating factor, it would be impossible to gather enough data on each individual address to create a statistically significant longevity model. The VitaCurves models first cluster addresses into ZIP+4 codes, postal codes or post codes, and then further using marketing definitions and statistical groupings to obtain robust models which have only a small number of geographical factors.