By Runhuan Feng and Siva Karthik Boddapati
Abstract
The undergraduate research program in risk and actuarial science at University of Illinois has been implemented for the past three years with an objective to provide students with opportunities to perform and disseminate research. Students are expected to develop skills in research methodology, project management, writing and presentation. This article summarizes the metrics for the achievements of program and experiences behind the planning, implementation and lessons gained from running the undergraduate research projects.
Introduction
The University of Illinois at Urbana-Champaign (Illinois) boasts one of the largest actuarial science programs in the nation with over 350 undergraduate students and around 40 graduate students as of spring 2017. The Illinois actuarial program provides a comprehensive actuarial education at all levels, including pre-professional undergraduate program, fast-track elite education at the master’s level, and a research-oriented Ph.D. concentration focusing on advanced analytics. It is one of the Centers of Actuarial Excellence (CAE) recognized by the Society of Actuaries and among the inaugural winners of the Casualty Actuarial Society’s University Award. The Illinois program is also highly regarded in the insurance industry for its rigorous curriculum preparing students for careers in all areas of actuarial practice. Built upon the past success of the program, we strive to provide our current students with the first-class education, stay updated with changing needs of the industry and bring in innovations in education.
As evidenced by numerous recent disruptive technologies to the insurance industry, such as autonomous vehicles, smart home technology, etc., there is a need for significant research and advancement with respect to identifying, quantifying and managing emerging risks that have little historical precedent. It is essential that future actuaries and risk analysts have the research skills necessary to engage with a complex and changing world, and to evaluate emerging future risks. The notion of an undergraduate research program in Risk and Actuarial Science was largely conceived by our former director, Dr. Richard Gorvett. It is intended to provide students with a platform to apply their knowledge in solving real-world problems. These projects serve as good out-of-class learning experiences by giving a preview of a professional work environment. The program was sponsored by a CAE education grant from the Society of Actuaries.
The benefits emanating from the undergraduate research program are numerous and profound:
- Development of an in-depth understanding of areas of potential importance for future risk and insurance activities.
- Hands-on experience with software programs like R, Python, and SAS which would become major career side advantages.
- Contributing to a broader actuarial and educational communities in several ways by collaborating and sharing all its findings publicly through an online research repository.
https://math.illinois.edu/academics/actuarial-science/undergraduate-research)
Undergraduate Research Article on Machine Learning—An Example
The undergraduate research project on machine learning is one of the many successful projects under this program. This project introduces undergraduate researchers to the world of machine learning in a self-learning mechanism. The following are the descriptions of various stages in the project and learnings from each stage.
Project Design & Structure
The undergraduate researchers were divided into three groups and each group was assigned to cover one of the topics in machine learning. Students started from learning basics through literature review and continued in the following lines:
- Project scope and deliverables:
- A comprehensive list of deliverables and the scope of the project was provided to the students. This helped students to pre-plan their work and contribute accordingly. The project scope defined the boundaries of the project and the list of deliverables helped students keep their progress in check.
- Project Timeline:
- Weekly reporting: All teams were expected to finalize a weekly meeting schedule with graduate student mentor and a bi-weekly schedule with faculty mentor.
- Mid-term review meeting: All teams were expected to present their work mid-way through the project including their documentation and code and obtain feedback from the mentors. This exercise proved to be rewarding as the groups were recalibrated in working towards the end goal.
- Final project submission: A final report submission date was announced at the inception of the project. All teams were expected to submit their final report a week before the last submission date as a final draft.
- Final report template:
- In addition, a template for the final project report was distributed to the students well in advance. This helped students to have a clear idea on how and what to deliver.
Data Procurement and Cleaning
Now-a-days, almost every project is data dependent, to implement machine learning algorithms and other models, real-world datasets must be procured from different sources. There are many data repositories that provide vast number of datasets that belong to various fields. The following are some the well-known data repositories:
In addition, students can also make use of governmental data sources such as United States Data and Statistics on www.usa.gov/statistics for specific data which fits their problem.
Moreover, the Casualty Actuarial Society with the help of CAS University Engagement Committee (UEC), initiated case competitions for university students and provided case toolkits comprised of case studies and other materials. In case of the machine learning project, each group identified a dataset that would fit their modelling requirements. Once the data was selected, a substantial amount of time was spent in data cleaning.
Model Building and Design
Each group had a different approach to data analysis and prediction. Once the datasets were chosen and cleaned, students attempted to model the data. The students implemented advanced machine-learning algorithms like decision trees, gradient descent etc., and analyzed the results.
Achievements
We have made significant achievements over the past three years with the support from the SOA and the CAE Grants program, and the Department of Mathematics.
- Large student participation: There have been between 30 and 50 students actively involved in the undergraduate research program each of the past three years. With between 80 and 100 undergraduate students graduating each year, engaging these many students in research would represent roughly half of our actuarial student population on average.
- Undergraduate research website: We have developed a website to promote undergraduate research in quantitative risk management and actuarial science. The website provides description of research projections, acts as a gateway for interested students to learn about and apply to program and repository for publicly available project reports and materials.
- Corporate engagement: Several companies participated in the program by suggesting research project ideas. Such engagement allows our students to learn about various topics in practical research.
- Research seminar: We have established Actuarial Science and Financial Mathematics seminar, which provides opportunities for students, faculty and outside research to present and discuss research projects and results.
- Creation of new courses: We have taught two topics courses for which research work from some projects have been used, one of which is a capstone course and the other focuses on the Modeling of Equity-Linked Insurance. With the success and gained experiences from project management, we expect to offer an actuarial research practicum, Risk Analytics and Decision Making, in spring 2018.
Conclusion
There have been many course-based undergraduate research projects in our actuarial program in the past. However, due to the time limit and required relevance of projects to class-room teaching, course-based research projects cannot offer the full research experience as open-ended problems do. Since the undergraduate research program has been implemented, we have seen growing interests from students and witnessed success for developing students’ research skills. From the early stages of student selection till the final report writing stage, careful monitoring and guidance is being provided to the students. A systematic approach towards students’ selection to the projects, which do not solely base on their academic background but also on their enthusiasm and commitment levels, is being followed.
So far, the feedback obtained from students was excellent. Students mentioned that this program provided them with both opportunities and challenges. We have learnt that upon successful completion of such projects, students are more confident in those areas of study and were ready to take up new challenges/responsibilities be it in research or in their career.
The program is aiming to identify the expectations of students, introduce undergraduate students to research practices and realities, provide a platform for those who are interested in research and academic careers, encouraging scientific inquiry and innovation among students and helping them develop time management, presentation and writing skills through a systematic approach. Several employers have expressed interest in students coming out of the undergraduate research program. Even though there is a lot of scope for the improvement of the program, we believe, with proper guidance and support this program can continue to grow and strengthen the reputation of our Illinois program as one that graduates students with advanced technical and strong communication skills.
Runhuan Feng, Ph.D., FSA, CERA, is director of actuarial science in the Department of Mathematics at the University of Illinois at Urbana-Champaign. He can be reached at rfeng@illinois.edu.
Siva Karthik Boddapati is lead graduate assistant in the Department of Mathematics at the University of Illinois at Urbana-Champaign.