490 – Recent Trends in Forecasting
Flow Field Forecasting: An Introduction
Sponsor: Section on Statistical Learning and Data Mining
Keywords: forecasting, Gaussian process regression, inductivism, penalized spline regression, time series
Kyle A. Caudle
South Dakota School of Mines and Technology
Michael Frey
Bucknell University
A statistical learning methodology, called flow field forecasting, is introduced for predicting the future of a univariate time series. Flow field forecasting draws information from the interpolated flow field of an observed time series to build a forecast step-by-step. Flow field forecasting is premised on the principle of inductivism. We present flow field forecasting, along with a discussion of its motivating principle, examples with comparisons to other major forecasting techniques and a statistical error analysis.
