Business Context
An insurance company wants to estimate the total payout for its auto insurance portfolio for the upcoming year. They model the frequency of claims using a Poisson distribution and the severity of each claim using a Log-Normal distribution. The goal is to generate a distribution of total annual losses.
Data Preparation
Creation of a severity definition table specifying a Log-Normal distribution with mu=8 and sigma=1.5.
Copied!
| 1 | |
| 2 | DATA mycas.sev_def_auto; |
| 3 | LENGTH model $8; |
| 4 | model='Logn'; |
| 5 | mu=8; |
| 6 | sigma=1.5; |
| 7 | dist='LOGN'; |
| 8 | OUTPUT; |
| 9 | |
| 10 | RUN; |
| 11 |
Étapes de réalisation
1
Execution of the CDM action with Poisson frequency and defined Log-Normal severity.
Copied!
| 1 | PROC CAS; |
| 2 | cdm.cdm / |
| 3 | seed=12345, |
| 4 | nReplicates=10000, |
| 5 | countDistributions={{name='POISSON', parmValues={{value=15}}}}, |
| 6 | severityDefinitions={name='mycas.sev_def_auto'}, |
| 7 | OUTPUT={outSample={name='auto_claims_sim', replace=true}, sampleVariable='TotalLoss'}, |
| 8 | outsum={outSummary={name='auto_claims_stats', replace=true}}; |
| 9 | RUN; QUIT; |
Expected Result
The action should successfully generate 10,000 replicate simulations of total annual losses. The 'auto_claims_sim' table will contain the individual simulation results, and 'auto_claims_stats' will provide summary statistics (mean, VaR, etc.) confirming the model integration.