Linear Prediction Filtering Programs:

Auto regressive (AR), moving average (MA), and auto regressive moving average (ARMA) filters are a method of representing the relation between inputs and outputs of a physical system by a linear model. These models are usually determined from historical data by solving a set of equations based on auto and cross correlation functions of the input and output. The models can also represent non-linear systems by using data dependent on the state of the system.

A major constraint on the use of these techniques in space physics has been the lack of continuous input on output time series. Another constraint is the rapid changes in the state of the system. Both problems disrupt the continuous time series needed to calculate correlation functions.

We have developed techniques based upon the theory for calculating these filters of multiple linear regression. These methods enable one to use any sequence of data which is as least as long as the filter to help define the filter. Both missing data and selection according to system state are easily accomplished.