Table of Contents
tabview - prototype dynamic query table viewer
tabview is an interactive program that displays
numeric data from an ASCII table in a 2D plot (see also tabplot(1NEMO)
while dynamically querying related columns from the same table. Dynamic
Query (DQ) is an effective way to visualize an N-dimensional dataset in
2D. Two variables are selected to be plotted, they can either be directly
taken as a column from the table, or computed via a fie(3NEMO)
involving any column, e.g. "sqrt(%1**2+%2**2)".
Each selected ’column’ can be
manipulated using sliders. Between the min (on the left) and max (on the
right) of a column, two sliders, called lo and hi can be moved around
independantly. lo is a lower limit, and hi the higher limit, for data to
be included in the display:
lo < hi MIN-------[lo=======hi]--------MAX
lo > hi MIN=======hi]--------[lo=======MAX
The lo slider can also be moved to the right of the hi slider, as the diagram
shows, in which case the sample will include data between min and hi as
well as lo and max. This is like swapping the lo and hi sliders and corresponds
to negating the logic of the first case.
Finally, only data are displayed
for which the corresponding slider value of all values falls in the selected
range. In SQL language, this means all queries are AND-ed.
Note: in this
prototype the words slider and column can be used interchangebly.
following parameters are recognized in any order if the keyword is also
Select a level of statistics output at startup of the program. At the lowest
level (0) an overview of the read data is given, including min, max, mean,
dispersion, skewness and kurtosis. At level 1 the Pearson correlation matrix
is given. At level 2 all possible linear regressions are given. At level
3 at multiple regressions are given. [Default: 1]
the final program will have (X-window) graphic sliders, the prototype has
a simple command line interface to select and manipulate sliders which
provide the dynamic query. Only the first character is matched, and for
some commands the second parameter is also used.
- Input table filename. No default.
- col=name1, name2,..
to be named and selected as sliders. The default column names are 1,2,3,...
for ascii tables. The default is that all columns are selected to become
- X-variable to plot. Any fie(3NEMO)
input columns (see col=) can be given. The first extracted column given
by col= is referred to as %1 etc. [Default: %1].
to plot (see xvar=). [Default: %2].
- Range in X from xmin
to xmax [autoscale + 5% edges].
- Range in Y from ymin to
ymax [autoscale + 5% edges].
- A fie(3NEMO)
which is used to colorize the particles. Color will be an integer, 0 being
invisible (actually: background color), 1 the foreground color, and 2,3,4...
the colors your graphics device has. [Default: 1].
of points in cm. This is allowed to be a fie(3NEMO)
expression, in terms
of other columns. It must evaluate in cm. Default: 0 (dots - this is also
the fastest and highly recommended) *** not implemented ***
- Point type. This is allowed to be a fie(3NEMO)
expression, in terms of
other columns. After evaluation, is is converted to an integer, see yapp(3NEMO)
for valid ptype’s. Default: 0 (a point) *** not implemented ***
- Filename with layout(5NEMO)
commands to be added to plot? Default: none.
- Whenever output is requested (the o command, see below),
it is written to this file. Warning: Any previously existing file will
be overwritten. Default: none.
- List of options, separated
by comma’s, that determine the output when out= has been selected. Valid
values are i: the ordinal (1 being the first in the original data) x,y:
the plotted data, data: the original (cols= only) data from the input table.
Note that the order is always i, x, y, data [Default: i,x,y,data]
Slider 2 [0.0387063 0.0387063 0.998341 0.998341]:
shows the slider name (normally numbers 1...nsliders), followed by the min,
lo, hi and max along the slider. The min and max can currently not be modified.
The following commands are recognized:
l <val> set lo slider to <val> 
h <val> set hi slider to <val> 
l s <step> set lo stepper to <step> 
h s <step> set hi stepper to <step> 
b <step> set both lo and hi stepper to <step> 
[0-9]* select a new slider by number (>0)
r force full redisplay
s show all sliders info
o (over)write output (needs out= keyword)
!cmd run (previous) unix command
|cmd run data through (previous) pipe
? this help
Noteworthy are the ! and | commands. They both have an optional cmd argument,
which allows the user to repeat a previously set cmd for this command.
The ! simply runs the cmd command through the shell, whereas the | pipes
all currently visible data into the cmd command. For example to do a linear
least squares fit to the visible data,
would be needed (see tablsqfit(1NEMO)
), all subsequents fits can be done
command itself, since it remembers the last used |cmd.
the screen is the dominant factor in the dynamic nature of tabview. On a
SPARC-1 a single average selection takes about 30ms for N=1000 rows and
M=4 columns. Updating a dummy yapp display took 200ms in this case, but
the dependance on N is highly nonlinear.
Recognize ASCII as well as
FITS tables (TABLE and BINTABLE). See also fitstab(1NEMO)
Need an improved
yapp interface - PGPLOT is clearly too slow. Graphical user interface.
number of rows, columns and linelength are defined in the program, no dynamic
memory allocation in this sense yet. Recompilation using -DMAXPOINTS, -DMAXSLIDERS
and -DMAXLINE resp. would be required.
Improve selection algorithm.
Tukey’s "Four Essentials" (c.1972)
with ack to: Human-Computer Interaction Laboratory - Ben Schneiderman et
8-oct-92 V0.0 prototyped PJT
11-oct-92 V0.1 added more commandline functionality PJT
2-dec-92 V0.2 added | and ! commands - corrected doc PJT
4-mar-94 V0.3 aded ’b’ command, more moment stats PJT
1-nov-95 V0.4 documented new features PJT
Table of Contents