Abstract
In this chapter, some of the ways of deriving the parameter coefficients of a model are discussed — a process I shall refer to as model construction, parameter estimation or modelling. Many different methods of model construction exist, and there are many theoretical arguments why one method is better than another, or why such and such a method is/is not appropriate in a given set of circumstances. However, for predictive modelling the method used to derive a model is arguably unimportant, as long as the final model is a good predictor of behaviour and satisfies any business requirements about the properties of the model. To some extent, the ends justify the means. There is no reason why you can’t just make up a model’s parameters based on your expert opinion, and in situations where data is scarce this is exactly what industry experts sometimes do. However, under normal conditions, where a good quality sample of data is available, modelling techniques that apply mechanical procedures (statistical or mathematical processes) nearly always generate models that are superior, in terms of predictive ability, than models created using more judgemental/subjective means.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Copyright information
© 2010 Steven Finlay
About this chapter
Cite this chapter
Finlay, S. (2010). Model Construction (Parameter Estimation). In: Credit Scoring, Response Modelling and Insurance Rating. Palgrave Macmillan, London. https://doi.org/10.1057/9780230298989_7
Download citation
DOI: https://doi.org/10.1057/9780230298989_7
Publisher Name: Palgrave Macmillan, London
Print ISBN: 978-1-349-36689-7
Online ISBN: 978-0-230-29898-9
eBook Packages: Palgrave Economics & Finance CollectionEconomics and Finance (R0)