raverage -- running average, standar deviation, and envelope
raverage input nwin
- Input one or two column list of numbers. Any line that can't be read as one or two numbers is ignored which means comments are allowed. The special name "STDIN" may be used to read the numbers from the standard input pipe or redirection.
- The number of values in the running average window.
- sort = no
- Numerically sort the first column of the input list by increasing value? This is done in an temporary file and the actual input file is not modified.
- nsig = 0
- The number of standard deviations below and above the average for the envelope columns. If the value is greater than zero two extra columns are output formed by subtracting and adding this number of standard deviations to the average value. If the value is zero then the columns not be written.
- fd1, fd2
- Internal parameters.
This task computes the running average and standard deviation of a one or two column list. For a one column list the ordinal value is added. Note that the ordinal is only for the lines that are successfully read so any comments are not counted. So the internal list is always two columns.
The input may be a physical file or the standard input. The standard input is specified by the special name "STDIN". All the input values are read and stored in a temporary file prior to computing the output. A temporary file is also used if the input is to be numerically sorted by increasing value of the first column. Note that the sorting is done before adding the implied ordinal for one column lists.
The output has four or six columns depending on whether nsig is zero or greater than zero.
average1 average1 stddev number [lower upper] average1 - the running average of the first column average2 - the running average of the second column stddev - standard deviation of the second column number - number of values in the statistic lower - optional lower envelope value upper - optional upper envelope value
The "number" of values may be less than the window if the window size is larger than the list.
The number of lines will generally be less than the input because there is no boundary extension. In other words the first output value is computed after the first nwin values have been read and the last output value is computed when the end of the list is reached.
The envelope columns are computed when nsig is greater than zero. The values are
lower = average2 - nsig * stddev upper = average2 + nsig * stddev
In many cases the data is intended to represent a scatter plot and one wants to show the trend and envelope as a function of the first column. This is where the sorting and envelope options are useful.
1. Compute the running average with a window of 100 values on the list of numbers in file "numbers".
cl> raverage numbers 100
2. Do this using the standard input. In this example use random numbers.
cl> urand 100 1 | raverage STDIN 90
3. Make a scatter plot of a two column list with the trend and envelope overplotted.
cl> fields numbers 1,3 | graph point+ cl> fields numbers 1,3 | raverage STDIN 100 sort+ nsig=3 > tmp cl> fields tmp 1,2 | graph append+ cl> fields tmp 1,5 | graph append+ cl> fields tmp 1,6 | graph append+