01.06.2015 Views

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Mixed Poisson Models for <strong>Claim</strong> Numbers 7<br />

as possible, the proportion <strong>of</strong> times that an event A occurs would behave according to the<br />

definition <strong>of</strong> Pr. Note that PrA is then the mathematical idealization <strong>of</strong> the proportion <strong>of</strong><br />

times A occurs.<br />

1.2.5 Independent Events<br />

Independence is a crucial concept in probability theory. It aims to formalize the intuitive<br />

notion <strong>of</strong> ‘not influencing each other’ for random events: we would like to give a precise<br />

meaning to the fact that the realization <strong>of</strong> an event does not decrease nor increase the<br />

probability that the other event occurs. Formally, two events A and B are said to be<br />

independent if the probability <strong>of</strong> their intersection equals the product <strong>of</strong> their respective<br />

probabilities, that is, if PrA ∩ B = PrAPrB.<br />

This definition is extended to more than two events as follows. The events in a family <br />

<strong>of</strong> events are independent if for every finite sequence A 1 A 2 A k <strong>of</strong> events in ,<br />

[ ]<br />

⋂ k k∏<br />

Pr A i = PrA i (1.1)<br />

i=1<br />

The concept <strong>of</strong> independence is very important in assigning probabilities to events. For<br />

instance, if two or more events are regarded as being physically independent, in the sense<br />

that the occurrence or nonoccurrence <strong>of</strong> some <strong>of</strong> them has no influence on the occurrence<br />

or nonoccurrence <strong>of</strong> the others, then this condition is translated into mathematical terms<br />

through the assignment <strong>of</strong> probabilities satisfying Equation (1.1).<br />

i=1<br />

1.2.6 Conditional Probability<br />

Independence is the exception rather than the rule. In any given experiment, it is <strong>of</strong>ten<br />

necessary to consider the probability <strong>of</strong> an event A when additional information about the<br />

outcome <strong>of</strong> the experiment has been obtained from the occurrence <strong>of</strong> some other event B.<br />

This corresponds to intuitive statements <strong>of</strong> the form ‘if B occurs then the probability <strong>of</strong> A is<br />

p’, where B can be ‘March is rainy’ and A ‘the claim frequency in motor insurance increases<br />

by 5 %’. This is called the conditional probability <strong>of</strong> A given B, and is formally defined as<br />

follows. If PrB > 0 then the conditional probability PrAB <strong>of</strong> A given B is defined to be<br />

PrAB =<br />

PrA ∩ B<br />

(1.2)<br />

PrB<br />

The definition <strong>of</strong> conditional probabilities through (1.2) is in line with empirical evidence.<br />

Repeating a given experiment a large number <strong>of</strong> times, PrAB is the mathematical<br />

idealization <strong>of</strong> the proportion <strong>of</strong> times A occurs in those experiments where B did occur,<br />

hence the ratio (1.2).<br />

It is easily seen that A and B are independent if, and only if,<br />

PrAB = PrAB = PrA (1.3)

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

Saved successfully!

Ooh no, something went wrong!