This repository contains small python projects that are based on actuarial topics. The purpose of this repository is for me to practice my coding skills, as well as to understand how probability, statistics and simulations can be used in actuarial work.
The purpose for this project is to show how an insurance company can use simulations to estimate future claim costs and decide how much it should charge per policyholder.
This project simulates the total insurance claims made by a group of policyholders over the period of one year. The number of claims is generated using a Poisson distribution, while the amount of each claim is generated using a normal distribution. By repeating the simulation many times, we can see how much the total annual loss may differ from year to year. The results are then used to estimate the average annual loss and to calculate a premium per policyholder.
The project also creates a histogram of the simulated annual losses. This makes it easier to see how much the insurer's costs can differ between years and why premiums should account for both the expected loss and the risk of higher than average claims.
The purpose of this project is to show how an actuary can estimate the current value of a company’s future pension promises. The project creates a fictional group of employees with different ages, salaries, years of employment and pension statuses. Employee ages are generated using a Weibull distribution, while salaries are generated using a lognormal distribution. The ranges and reference values are based mainly on Dutch data, together with several simplifying modelling assumptions.
For each employee, the project calculates the annual pension benefit based on their salary, years of employment and the chosen pension accrual rate. It then estimates how much their future pension payments are worth today by accounting for the time until retirement, survival probabilities and the discount rate. Finally, the results are combined to calculate the company’s total pension liabilities and its expected pension payments for each future year. This shows how employee characteristics, survival assumptions and discounting can affect the value and timing of a company’s pension obligations.
The purpose for this project is to show how an insurance company can use information about a driver to estimate their expected claim costs, and decide how much it should charge them for car insurance.
In this project I created fictional drivers with different ages, years of driving experience, annual mileages, previous accidents, and car values. Driver ages and licence ages are generated using the Beta distributions, while annual mileage and car value are generated using lognormal distributions. The number of previous accidents is generated using the Poisson distribution. The ranges and reference values are based partially on Dutch data, together with several simplyfying modelling assumptions.
For each policyholder, the project estimates the expected claim frequency based on their previous accidents, annual mileage and driving experience. It then estimates the average cost of one claim, which is partly dependent on the value of the car. The expected claim frequency and average claim cost are combined to calculate the expected annual loss. Finally, an additional percentage is added for expenses and profit to calculate an annual premium per policyholder. This shows how the characteristics of a driver and their car can affect the premium charged by an insurance company.