![IconGems-Print](images/IconGems-Print.png)
547 – New Developments in Machine Learning
Next Generation Flow Field Forecasting
Kyle Caudle
South Dakota School of Mines and Technology
Patrick Fleming
South Dakota School of Mines and Technology
Michael Frey
Bucknell University
Noah Brubaker
South Dakota School of Mines and Technology
Flow field forecasting was first developed in 2011 as a method to forecast a univariate time series. The original version of flow field forecasting which is available on the Comprehensive R Archive Network (CRAN) was shown to be a competitive alternative to Box-Jenkins ARIMA, exponential smoothing and neural networks. Flow field forecasting has several very nice features such as, (1) reduction of historical archived data, (2) autonomous operation, and (3) computational efficiency. This talk will focus on the next version of flow field forecasting which will forecast a bivariate response (e.g. latitude, longitude). Other advancements to be touched on will be the inclusion of external environmental factors such as weather and geography.