MRMR: Multivariate Regression Models for Reserving

Description Details


MRMR allows an actuary to create sets of loss data and forecast liabilities. It uses a set of 3 S4 objects to store data, models and predictions.



A Triangle is a collection of aggregate loss data. All triangles must have a defined set of OriginPeriods, a defined set of DevelopmentIntervals and data along those axes. A triangle may carry additional descriptive information such as line of business, geographic region and so on.


A TriangleModel is a statistical model fit to triangle data. The formula may be defined by the user and will generally be a linear or generalized linear model. A triangle may have more than one model. It usually will.


A TriangleProjection is a prediction based on a TriangleModel. A TriangleModel may have more than one projection.

Search within the MRMR package
Search all R packages, documentation and source code

Questions? Problems? Suggestions? or email at

Please suggest features or report bugs with the GitHub issue tracker.

All documentation is copyright its authors; we didn't write any of that.