*The Dow Company*
**Proposed Project for the
MSU Industrial Math Students**

Using Neural Networks for High Frequency Forcasting

**Project Overview:** Forecasting with high-frequency data
(i.e., data that is daily or more frequent) presents special challenges for
analysts. It has been found (e.g., in Crone (2009)) that neural networks
sometimes present a good answer to these challenges. For this project we are
looking at an evaluation of neural networks as a possible tool for different
types of high frequency time series data analysis. The first part of this
project will be to attempt to use neural networks as an aid in forecasting for a
variety of different high frequency time series. The second part will be to
determine whether there are certain qualitative characteristics of these time
series which might predict which time series in general might be amenable to
modeling by neural networks.