MOVING STATISTIC PLOT

... MOVING STATISTIC PLOT

Graphics Command

Plots the moving value of a statistic for a response variable. Optionally, the moving statistic can be plotted against a group-id variable.

It is sometimes useful to a compute a statistic over a "local" region of a variable rather than the entire variable. The most common example would be to compute a moving average for a variable.

Although this command is most commonly used for a location statistic, it can in fact be used for any supported statistic (enter HELP STATISTIC for a list of supported statistics).

The local area is defined by the FILTER WIDTH command. For example,

FILTER WIDTH 5

says that 5 points will be used to compute the specified statistic.

The command

SET MOVING DIRECTION <CENTER/LEFT/RIGHT>

defines the direction of the local region:

CENTER	=	the current row plus an equal number of rows above and below the current row. For example, if the FILTER WIDTH is set to 5, the local area for row i will be rows i - 2 to i + 2.
LEFT	=	the current row plus the specified number of rows above the current row. For example, if the FILTER WIDTH is set to 5, the local area for row i will be rows i to i + 4.
RIGHT	=	the current row minus the specified number of rows below the current row. For example, if the FILTER WIDTH is set to 5, the local area for row i will be rows i - 4 to i.

The other issue is how the end points will be handled. This is specified with the command

SET MOVING END POINT <SKIP/SYMMETRIC/PARTIAL>

where

SKIP	=	if the local region extends beyond the start or end point, no output value will be generated for that row.
PARTIAL	=	if the local region extends beyond the start or end point, the statistic will be computed for the subset of rows that are within range.
SYMMETRIC	=	if the local region extends beyond the start or end point, the statistic will be computed for the largest subset of rows that maintains an equal number of rows above and below the current row. This option is only supported if the MOVING DIRECTION is CENTER.

The default is a FILTER WIDTH of 3, MOVING DIRECTION of CENTER, and a MOVING END POINT of SKIP.

The <stat> STATISTIC PLOT can be used plot the value of a statistic versus the index of a group-id variable. So if you have 10 groups in your data, there will be 10 values of the statistic computed.

The <stat> MOVING STATISTIC PLOT is a variant of the STATISTIC PLOT.

There are two cases.

1. If there is a single group in the data, then the moving value of the statistic will be plotted versus the sequence number. That is, plot
   \( S_{Y_i} \) versus i

   where

   \( S_{Y_i} \) versus i = the value of the moving statistic at point i.

   A reference line will be drawn at the value of the statistic for the full data set.

   The appearance of these two traces is controlled by the first two settings of the LINES, CHARACTERS, SPIKES, BARS, and and associated attribute setting commands.

2. If there are multiple groups in the data, then the process above is repeated for each group. In this case, the x-coordinates is the index of the full data set, not just the current group. However, the moving statistic is only computed for points in the current group. The group sizes do not need to be equal.

   The distinction between this command and the STATISTIC PLOT is that the STATISTIC PLOT only plots the value of the statistic for all points in the group while this command plots the value of the moving statistic for all points in the group. For example, if there are ten groups with ten points each, the STATISTIC PLOT will plot ten points while the MOVING STATISTIC PLOT will plot 100 points.

   A reference line will be drawn for each group (the reference value is the value of the statistic for all points in the specific group).

   The traces for the plot are defined in pairs. That is, trace one is the moving statistics for group one, trace two is the reference line for group one, trace three is the moving statistics for group two, trace four is the reference line for group two, and so on.

   If you enter the command

   SET MOVING STATISTIC PLOT GROUPS STACKED

   then the x coordinate will start at one for each group. In this case, no reference line is drawn.

   To turn off the stacking option, enter

   SET MOVING STATISTIC PLOT GROUPS DEFAULT

<stat> MOVING STATISTIC PLOT <y1> ... <yk>
                        <SUBSET/EXCEPT/FOR qualification>
where <stat> is one of Dataplot's supported statistics;
            <y1> ... <yk> is a list of 1 to 3 response variables (<stat> determines how many response variables);
and where the <SUBSET/EXCEPT/FOR qualification> is optional.

For a list of supported statistics, enter

HELP STATISTICS

This syntax is for the case without groups in the data.

<stat> MOVING STATISTIC PLOT <y1> ... <yk> <x>
                        <SUBSET/EXCEPT/FOR qualification>
where <stat> is one of Dataplot's supported statistics;
            <y1> ... <yk> is a list of 1 to 3 response variables (<stat> determines how many response variables);
            <x> is a group-id variable;
and where the <SUBSET/EXCEPT/FOR qualification> is optional.

For a list of supported statistics, enter

HELP STATISTICS

This syntax is used for the case where there is a group-id variable.

MEAN MOVING STATISTIC PLOT Y
MEAN MOVING STATISTIC PLOT Y X
MEAN MOVING STATISTIC PLOT Y X SUBSET X > 2
SD MOVING STATISTIC PLOT Y
CORRELATION MOVING STATISTIC PLOT Y1 Y2

If the command
SET MOVING END POINT SKIP

is given, then some points at the beginning or end of the data will not have the moving statistic computed. So if i=3 is the first point at which the moving statistic is computed, the x-coordinate will be set to 3, not 1.

The word STATISTIC is required in this command (i.e., MOVING PLOT is not a synonym for MOVING STATISTIC PLOT). This is to avoid conflicts with other commands.

None

None

MOVING <STATISTIC>	= Compute the moving value of a statistic.
CUMULATIVE STATISTIC PLOT	= Generate a plot of the cumulative value of a statistic.
SMOOTH	= Perform a smooth.
STATISTIC PLOT	= Generate a statistic versus index plot.
CHARACTERS	= Sets the type for plot characters.
LINES	= Sets the type for plot lines.

Exploratory Data Analysis

2015/5

The list of supported statistics is frequently updated. Enter HELP STATISTICS for a current list of supported statistics.

 
. Step 1:   Read the data
.
dimension 40 columns
skip 25
read zarr13.dat y1
read berger1.dat y2 x2
read lew.dat y3
read gear.dat y4 x4
skip 0
.
. Step 2:   Set some default plot control options
.
case asis
label case asis
tic mark label case asis
title case asis
title offset 2
y1label displacement 15
x1label displacement 12
multiplot scale factor 2
multiplot corner coordinates 5 5 95 95
.
. Step 3:   Moving plots, no groups, single response variable
.
filter width 5
x1label Sequence
.
multiplot 2 2
.
title Mean Moving Statistic Plot
y1label Moving Mean
mean moving statistic plot y1
.
title Median Moving Statistic Plot
y1label Moving Median
median moving statistic plot y1
.
title SD Moving Statistic Plot
y1label Moving SD
standard deviation moving statistic plot y1
.
title Skewness Moving Statistic Plot
y1label Moving Skewness
skewness moving statistic plot y1
.
end of multiplot
.
just center
move 50 97
text Moving Statistic Plots for ZARR13.DAT (filter width 5)
.
y1label displacement
x1label displacement
.
. Step 4:   Moving plots, no groups, two response variable
.
xlimits 0 120
.
title Correlation Moving Statistic Plot for BERGER1.DAT
y1label Moving Correlation
correlation moving statistic plot y2 x2
.
xlimits
.
. Step 5:   Cumulative/moving plots, groups, one response variable
.
title Mean Moving Statistic Plot for GEAR.DAT
y1label Moving Mean
mean moving statistic plot y4 x4