Industrial Projects

2011-2012 Projects

Evaluating the Robustness and Feasibility of Integer Programming and Dynamic Programming in Aircraft Sequencing Optimization

Sponsor: MIT Lincoln Laboratory
Advisors: Jon Abraham, George Heineman (CS)
Students: Julia Baum, William Hawkins

This project uses Mixed Integer Linear Programming and Dynamic Programming to optimize the takeoff sequence of aircraft at Dallas/Fort Worth (DFW) by minimizing departure delay while also obeying separation requirements and position shifting constraints. We modeled taxi time uncertainties based on real data from DFW and analyzed the robustness of the optimization solution and the feasibility of using these methods in real-life. The runtimes of these methods proved to be feasible in real-time, however the solutions failed to be robust, creating a future need for a stochastic optimization.

A Process for Analyzing Change and Its Application to Commercial Truck Insurance

Sponsor: Travelers Insurance
Advisor: Jon Abraham
Students: Alana Aubin, Benjamin Blakeslee, Monique Cote, Ryan Landry

In business, it is important to be able to adapt to changes quickly and effectively. Government regulations are frequently changing and affect many different industries of American business. This paper outlines a process for evaluating and responding to change, with an in-depth example of its application to proposed hours of service regulations for truck drivers, and how new safety regulations will impact commercial auto insurers. Ultimately, we find the process to be effective in quantifying the impact of this legislation on expected future costs for the insurer. More broadly, the process outlined in the paper gives a structured approach to responding to change and can be used in a variety of situations.

Global Credit Products Volume and Market Share Reporting

Sponsor: Bank of America Merrill Lynch
Advisors: Jon Abraham, Tsung-Yi Wang (MG)
Students: Hui Cheng, Lauren Moloney

One of the main goals for Global Credit Trading is the ability to aggregate Cash and CDS data and store all daily transactions. In addition, volume has become more central to performance analysis. An accurate picture of trade volume is frequently requested and important to the overall evaluation of the trading desk. The current process for volume and market share aggregation is fragmented over many systems and run independently under different methodologies in all regions. The Credit Trading Strategists have requested the creation of internal trade history database. The goal of this project was to improve the manual procedure by providing data analysis on the current reporting system and proposed solutions to build centralized internal database to store both Cash and CDS data.

Tactical Data Management

Sponsor: BNP Paribas
Advisors: Jon Abraham, Arthur Gerstenfeld (MG)
Student: Christopher Pryor

The project focused on delivering a business intelligence system to BNP Paribas that would store data for easy access by the end users, while providing important auditing information for the business side. The solution integrated Microsoft Excel, Tableau, a Python module, and a standalone CSV module that could each communicate with a central database. By normalizing each data item, the data was independent of the originating program and could be successfully imported by any of the appropriate parts of the system. The intent is that this system will help to facilitate collaboration within the organization, reduce the time spent on data intensive tasks, and provide important records for compliance purposes.

BNP Paribas: Enterprise Architecture

Sponsor: BNP Paribas
Advisors: Jon Abraham, Daniel Dougherty (CS)
Student: Joseph Servi

The goal of the project, sponsored by the Global Equities and Commodity Derivatives (GECD) at BNP Paribas, was to create a graphical management application that allows the group to better visualize a wide array of system flows. It was required for the application to have bi-directional communication with internal BNP databases and clearly show any possible problems with the data, such as an overloaded server. The outcome of the project is an integrated application that graphically displays and interactively manages business flows and their underlying data and builds performance/capacity dashboards.

Predictive Loss Ratio Modeling with credit scores, for insurance purposes

Sponsor: Hanover Insurance Group
Advisor: Jon Abraham
Students: Corey Alfieri, Taylor Ketterer, Ricardo Obasare, Kyaw Thiha

In recent times, credit scores have gained widespread popularity within the insurance industry, especially in the underwriting and pricing, due to their powerful predictive value. The goal of this project was to create a loss ratio model that would improve the predictive ability of the current Hanover premium model through implementation of credit scores. This would enable Hanover to benefit from more informed underwriting and pricing techniques, greater competitive advantage in commercial insurance lines of business and most importantly, more robust underwriting profit. Hanover's ability to better differentiate the risk types of their customers will ultimately improve their underwriting profit by ensuring that they do not underwrite policies with excessively high-risk.

Pigskin Party: A Statistical Analysis on Fantasy Football and “The Machine”

Sponsor: Advanced Sports Logic
Advisor: Jon Abraham
Students: Mark Johnston, Ari Lathrop, Nicholas Mondor

This project used a variety of different mathematical techniques to improve upon Advanced Sports Logic's fantasy football software product known as 'The Machine.' The team looked at the mathematics behind some of the functions used within the software and recommended changes accordingly. Additionally, the team also worked on creating a new product within 'The Machine' which projects statistics throughout the course of a season. The team concluded that the contents of this project could be expanded upon and recommended how to do so consequently.

Numerical Investigation for Variable Universal Life

Sponsor: John Hancock
Advisor: Jon Abraham
Students: Maarja-Liisa Sokk, Han Yang

During the sales process of an insurance product an illustration is used to show a customer the potential benefits and cash values of a policy. Our project was to determine what annual rate of return John Hancock should be using to illustrate their Indexed Universal Life policy at. To accomplish this we built a model for projected stock market returns using the Random Walk and Regime Switching models as a basis. We also layered on certain policy features to see how they affected the annual rate of return.

Maintained by webmaster@wpi.edu
Last modified: Feb 05, 2016, 14:04 UTC