University of New Hampshire
Mentor: Dr. Mark Lyon, UNH Department of Mathematics
Application of Fast Randomized Algorithms to Predictive Modeling
The Information Age has come with many challenges and obligations, whether they are perceived or imposed upon by society, industry or individuals themselves. One of these is the use of predictive models by industry to empower decision making by leaders. The Heritage Health Prize is one of the new age contests in large scale collaborative challenges funded by Kaggle, a data mining competition company. The challenge and necessity for the contest is highlighted by their observation below.
“More than 71 Million individuals in the United States are admitted to hospitals each year, according to the latest survey from the American Hospital Association. Studies have concluded that in 2006 well over $30 billion was spent on unnecessary hospital admissions. Each of these unnecessary admissions took away one hospital bed from someone else who needed it more.”
The contest is geared towards generating effective and precise algorithms for predicting the numbers of days in specific years given limited medical claims data. The study will use contest as an opportunity to explore the how recent developments in the application of certain dimensional reduction techniques can be applied.
In particular I hope to investigate how recent work in randomized algorithms for low-rank matrix approximations can be used. The study will be accomplished through an analysis of time and error rates, as well and additional quantitative variables. These measurements will be gathered from a range of different predictive algorithms implementation, both with and without the new randomized algorithms.