Claims estimation for Non-Life Insurance Using MATLAB

Event Type
Web Session
European Actuarial Academy

Start time: 9.00 am GMT

End time: 1.00 pm GMT


Announcement from the European Actuarial Academy organiser: How can you extract meaningful insights and models from the vast amounts of data now avilable, in a transparent, efficient and automated manner? How can you remain compliant with ever increasing regulatory demands? Fast, powerful and reliable tools are a necessity. These will allow us to quickly test traditional algorithms as well as newer techniques and achieve new insights. We’ll be able to focus our efforts on the important issues rather than tedious data cleaning, out of memory problems for big data or repetitive report generation. During these two sessions we’ll focus on modelling Claims estimation in an efficient and reusable manner. We’ll be modelling Claims estimation using different traditional methods, automating the whole workflow we encounter in our daily tasks: from data handling, modelling and report generation (static or dynamic) as well as automatic component creation to be run by anyone, anywhere via desktop or web. We’ll also see alternative methods to be applied to this kind of problem as well as how to handle data which doesn´t fit in memory without compromising any accuracy.

The web session is open to all interested persons.

Technical requirements: Please check with your IT department if your firewall and computer settings support web participation (the programme GotoTraining/GotoWebinar is used for the web session).

The registration fee is € 200.00 plus 16% VAT. Click here to make a reservation.


9.00 am - 11.00 am:

  • Importing, cleaning and visualizing data
  • Creating a development triangle and calculating IBNR and unpaid claims using chain ladder, expected claims and Bornhuetter-Ferguson methods.
  • Automated report generation (static or dynamic to a dashboard)


1.00 pm - 3.00 pm:

Building a model to estimate total settlement amount based on a number of predictors. We’ll be using multiple linear regression, a bag of decision trees and a neural network.


  • How to automate the whole workflow
  • How to handle big data which doesn´t fit in memory
Paula Poza
Biographical details

Paula is an Application Engineer at MathWorks focusing on financial applications, with more than 10 years of experience helping entities all around Europe. Paula has a Maths degree by the Universidad Complutense de Madrid and subsequently carried out actuarial studies at the Institute of Actuaries in London. Her professional career before MathWorks developed in Consultancies and financial entities in UK and Spain.