MATLAB Routines for Moving Median with Trend and Seasonality for Time Series Prediction

Electronic copy available at: http://ssrn. com/abstract=1522944 MATLAB Routines for Moving Median with Trend and Seasonality for Time Series Prediction Eleftherios Giovanis Abstract In this paper we provide a simple MATLAB routine which computes the... More

Electronic copy available at: http://ssrn.
com/abstract=1522944
MATLAB Routines for Moving Median with Trend and Seasonality for Time
Series Prediction
Eleftherios Giovanis
Abstract
In this paper we provide a simple MATLAB routine which computes the moving
median with trend and seasonality.
This approach is linear and for this reason has its
disadvantages.
So this routine can be improved by combining Monte-Carlo
simulations, genetic algorithms simulations and wavelets decomposition.

Keywords: moving median, MATLAB
Introduction
We present three MATLAB routines which replicate the moving median with
trend and seasonality with hourly data, 4-hourly data and daily data.
These routines
can be modified in a way to take data of higher frequency, as 15-minute or 30-minute,
as low frequency data, like weekly, monthly and quarterly.
But, in order to estimate
data of different frequency, modifications have to be made in the settings each time.

We just present below th Less