01.08.2014 Views

Estimation, Evaluation, and Selection of Actuarial Models

Estimation, Evaluation, and Selection of Actuarial Models

Estimation, Evaluation, and Selection of Actuarial Models

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Chapter 2<br />

Model estimation<br />

2.1 Introduction<br />

In this Chapter it is assumed that the type <strong>of</strong> model is known, but not the full description <strong>of</strong> the<br />

model. In the Part 3 Note, models were divided into two types–data-dependent <strong>and</strong> parametric.<br />

The definitions are repeated below.<br />

Definition 2.1 A data-dependent distribution is at least as complex as the data or knowledge<br />

that produced it <strong>and</strong> the number <strong>of</strong> “parameters” increases as the number <strong>of</strong> data points or amount<br />

<strong>of</strong> knowledge increases.<br />

Definition 2.2 A parametric distribution is a set <strong>of</strong> distribution functions, each member <strong>of</strong><br />

which is determined by specifying one or more values called parameters. Thenumber<strong>of</strong>parameters<br />

is fixed <strong>and</strong> finite.<br />

For this Note, only two data-dependent distributions will be considered. They depend on the<br />

data in similar ways. The simplest definitions for the two types considered appear below.<br />

Definition 2.3 The empirical distribution is obtained by assigning probability 1/n to each data<br />

point.<br />

Definition 2.4 A kernel smoothed distribution is obtained by replacing each data point with a<br />

continuous r<strong>and</strong>om variable <strong>and</strong> then assigning probability 1/n to each such r<strong>and</strong>om variable. The<br />

r<strong>and</strong>om variables used must be identical except for a location or scale change that is related to its<br />

associated data point.<br />

Note that the empirical distribution is a special type <strong>of</strong> kernel smoothed distribution in which<br />

the r<strong>and</strong>om variable assigns probability one to the data point. The following Sections will introduce<br />

an alternative to the empirical distribution that is similar in spirit, but produces different numbers.<br />

They will also show how the definition can be modified to account for data that have been altered<br />

through censoring <strong>and</strong> truncation (to be defined later). With regard to kernel smoothing, there are<br />

several distributions that could be used, a few <strong>of</strong> which are introduced in this Chapter.<br />

A large number <strong>of</strong> parametric distributions have been encountered in previous readings. The<br />

issue for this Chapter is to learn a variety <strong>of</strong> techniques for using data to estimate the values <strong>of</strong> the<br />

distribution’s parameters.<br />

3

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!