NSF Award
9461909
This Small Business Innovation Research Phase I project proposes research necessary to create an interactive, exploratory software environment for detecting and modeling change points. The problem of detecting and modeling change points is faced in many contexts in which more than one model is needed to describe the data. For example, health variables describing a patient may change drastically after heart surgery. The challenge is to detect change points, and to characterize the nature of the change. Phase I research will focus on (1) algorithms for robust methods, (2) interactive, dynamic graphics, (3) the state space approach to modeling seasonal data which contains change points, and (4) investigating the MTD model in which it is not necessary to specify in advance the presence of change points. Phase II will expand upon Phase I work, investigate methods for a broader set of problems, and develop a prototype implementation. Phase III will develop a final commercial product, implemented in the modern statistical package S-PLUS. This important set of modern statistical tools will enable applied statisticians, scientists, engineers, and other data analysts to fit more realistic models, and thereby realize the greatest return on their investment of collecting data. Such tools will help speed technological innovation, and make scientific research more efficient.
