This course will touch upon predictive analytics and modeling in life insurance – where it is used, the applications of predictive analytics and modeling in business. It also explains how to build a predictive model – data management, the types of predictive models, mortality models and other insurance applications. At the end, we will explain the results, ethics and legal limitations.
Link to course:
https://www.experfy.com/training/courses/predictive-analytics-and-modeling-in-product-pricing-personal-and-commercial-insurance
3. About Life Insurance
• Contract between an insurance policy holder and an insurer or
assurer, where the insurer promises to pay a designated beneficiary
a sum of money (the benefit) in exchange for a premium, upon the
death of an insured person (often the policy holder).
• Two categories: protection and investment policies.
• History: burial clubs in Rome, then in London 1706 there was first
company in London to offer life insurance.
4. • Why is it useful for individuals to have life insurance?
Individual Perspective
o Providing survivors way to pay monthly bills. Also, payment of
descendant's final expenses and mortgages.
o In USA, funding future education expenses.
o Paying amount due for assets.
o Insuring insurability. That is the chance your health gets worse soon.
5. • Insurance companies try to charge a premium and hopefully end
payment will be lower than sum of premiums collected.
• Insurance company takes on risk made by individual. By insuring
large pool of individuals, it diversifies risk and makes it manageable.
• Insurance company has fixed cost (administration and similar) and
also final claim.
• Premiums cover insurance company’s costs and also provide it with
profit.
Benefits for Business – Insurance
Companies
6. • There are legal problems, which we will not discuss in detail in this
course.
• Interesting for us: managing risk. Offering competitive premium
which will still provide insurance company a profit.
• First attempts (still relevant): mortality tables.
• Get the actual duration before person dies as close as possible.
• Special cases: What happens if there is a suicide?
Challenges for Insurance Company
o It gets complicated, but if it is within two years, then, in USA company
does not have to cover. This is different in different countries and even
states.
7. • Defining the best possible premium – amount of money insurance
policy holder pays and how often (usually every month).
• Defining benefits.
• What to do with paid premiums/money? If it just keeps it, inflation
might devalue it.
• Reinsurance problem – what if it comes to shock because of which a
lot of people die at once (example disease)?
• Assurance and insurance: also protection of events that might
happen
Challenges for Insurance Company
8. • Problems are usually of mathematical nature
• Insurance company needs models to reduce risk and predict the
future.
• This helps them with planning and making their business
sustainable.
• There comes the need for predictive models and modeling in
general.
Challenges for Insurance Company
9. Predictive Modeling Usability
• The mortality tables provide a baseline for the cost of insurance, but
the health and family history of the individual applicant is also taken
into account
• The investigation and resulting evaluation is termed underwriting.
• Factors to include:
o Personal medical history
o Family medical history
o Driving record
o Height and weight matrix, BMI (Body Mass Index)
10. Predictive Modeling Usability
• Predictive modeling also improves
marketing and risk classification.
• Fraud detection and differentiation is
another usage.
• Table: potential savings of
representative life insurer by using
predictive modeling.
Source: Batty et al (2010), Predictive modeling for life
insurance, Deloitte consulting LLP
11. Brief History of Predictive Analytics in
Insurance
• 17th century: use of mortality tables (John Graunt and Edmund Halley).
• In 20th century, general insurance actuaries used generalized linear
models and empirical Bayes techniques for pricing of short-term
insurance policies.
• Between 1787 and 1837 more than two dozen life insurance
companies were started, but fewer than half a dozen survived.
• As the time went on, the datasets were becoming bigger and more
things could be used for predictions.
12. Brief History of Predictive Analytics in
Insurance
• The real growth of predictive analytics came in last 50 years.
• They started using questioners and collecting people’s information
about their habits and their past records.
• https://www.geico.com/life-insurance/ - online insurance where
premium is immediately calculated.
• Above uses a lot of different variables.