Time Series Analysis Demo

The data shown in this demonstration is from feeder
circuits at a large power generation station.  The data
was collected from many circuits, every half-hour for
a period exceeding one year.

To display the data, first select a circuit by clicking
on the corresponding button.   Once a circuit has been
selected, you may, if desired, select a specific month
and/or day to subset the data.

NOTE: You will need to select a month before you
      can select a particular day, and you must
      select a circuit before you can select a
      month, since not all circuits have readings
      for the same dates.

Once the circuit has been selected, and any desired
subset has been chosen, you can view the data.
When viewed, the data will be processed by the
selected process.

This demo allows you to view the data using three
different methods:

1.  CHART

      The selected data is displayed as a line graph.
      If the process is "None" then the line color will
      be yellow, and you will be allowed to click on the
      graph.  The closest data point will be found, and
      the date, time, and data value of that point will
      be displayed.

      If the process type is "Least Absolute Deviation",
      Then the original data is plotted with red markers,
      and the data trend is displayed as a green line.
      The calculations are performed by IDL's "LADFIT"
      function.

      If the process type is "Autoregressive Forecast",
      then the original data is displayed as a red line,
      and the forecast data as a dotted green line.
      The forecast calculations are performed by IDL's
      "TS_FCAST" function.

      If a particular day is selected, and one of the
      24-hour processes is chosen, then the result of the
      process is a single value, which is displayed in a
      status window rather than as a graph.

2.  IMAGE

      The day must be set to "All" to display an image.
      Also, the process type must not be set to
      "Least Absolute Deviation".

      The selected data is displayed as an image with the
      day of the year (or month) on the X axis, and the time
      of day on the Y axis.   This is performed by IDL's
      "REFORM" function, which changes the data from a one-
      dimensional array of N elements to a two-dimensional
      array of dimensions M by 48, where M = N / 48.
      Note that there are 48 samples taken during each day
      (every half-hour).

      The "Autoregressive Forecast" process will display
      only the forecast data as an image.

      Once the image is displayed, you may drag the cursor
      over the image data, and this will cause dynamic
      profiles of the data to be displayed as graphs.
      The data value under the cursor will be displayed
      in the status area in the lower left corner of the
      image window.

3.  SURFACE

      The process type must be set to "None" to display a
      surface.   Also, the month must be set to "All".

      This method allows you to compare multiple data sets
      by using one set of data as the surface elevation,
      and a different set of data for the surface shading.