Cruise Scientific                Visual Statistics Studio                Visual Statistics Studio Guide

Time Series Analysis

Moving Averages

Fast Fourier Transforms

 

 

Moving Averages

Select (Project, New Project) and (Project, Open Project Files) commands

Click on the down arrow of the Files of type: selection box which opens as

Files of type:  
  Project Files (*.project)
  Project Files (*.correlational)
  Project Files (*.experimental)
  Project Files (*.factor)
  Project Files (*.logical)
  Project Files (*.longitudinal)
  Project Files (*.stereograms)

 

Select the Project Files (*.longitudinal) dialog

and click on the Movies (1910 -1968).longitudinal data file name

 

 
   

 Select (Analysis III,  Moving Averages) command

and click on the Accept command.

 

To understand the truncating of the mAvg vector, enter variable X [1 2 3 4 5 6 7] and compute its moving averages of order 1 (with default weights; also note that the Scale and the Trend can be the same variables)

 

Increase the order of the moving average up to the point when its order equals n, as

When the order of the moving average equals one, the moving average simply replicates the extant trend. When the order equals n, the moving average equals to the arithmetic mean. For values of the order of the moving average in-between of these marginal values, the length of the moving average vector, as compared with the length of its scale, shrinks, as the order of the moving average increases. To plot the values of the moving average, remove the top and bottom blank values by the ( Stratify, Select Cases) command

 and select (Graphs, Spline Graphs; indexed by entities)

The time series analysis described above pertains to movie production in a Central European country which is of interest as its market economy of the pre-World War II years was succeeded by the planed economy during the 1948 - 1989 time interval. Click on the magnifying glass on the graph's toolbar and extract a two 15-year intervals: the 1920 - 1935 interval spanning the years of the market economy, and the 1950 -1965 interval for the segment of the period of the planned economy. While the 1920 - 1935 trend shows marked fluctuations, the trend for the1950 - 1965 time period shows moderate, but steady growth.

 

The 1920 - 1935 period of market economy

The 1950 - 1965 period of planned economy

 

 

Fast Fourier Transforms

Using steps similar to those described above, we can apply the Fast Fourier Transforms (Analysis III, Fast Fourier Transforms) to the same data, using the same order parameter for both methods,

 

 and compare both trends:

The obtained trends are similar, however Fourier transformations for smoothing secular trends do not employ weights and do not leave blank intervals on the beginning and end of the trends analyzed.