Moving Average Trend Estimation
This example shows how to estimate long-term trend using a symmetric moving average function. This is a convolution that you can implement using conv. The time series is monthly international airline passenger counts from 1949 to 1960.
Load the airline data set (Data_Airline).
load('Data_Airline.mat') y = log(Data); T = length(y); figure plot(y) h = gca; h.XLim = [0,T]; h.XTick = [1:12:T]; h.XTickLabel = datestr(dates(1:12:T),10); title 'Log Airline Passenger Counts'; hold on
The data shows a linear trend and a seasonal component with periodicity 12.
The periodicity of the data is monthly, so a 13-term moving average is a reasonable choice for estimating the long-term trend. Use weight 1/24 for the first and last terms, and weight 1/12 for the interior terms. Add the moving average trend estimate to the observed time series plot.
wts = [1/24;repmat(1/12,11,1);1/24]; yS = conv(y,wts,'valid'); h = plot(7:T-6,yS,'r','LineWidth',2); legend(h,'13-Term Moving Average') hold off
When you use the shape parameter 'valid' in the call to conv, observations at the beginning and end of the series are lost. Here, the moving average has window length 13, so the first and last 6 observations do not have smoothed values.
See Also
Related Examples
- Seasonal Adjustment Using a Stable Seasonal Filter
- Seasonal Adjustment Using S(n,m) Seasonal Filters
- Parametric Trend Estimation