3. SGI - Statistical Graphics Interface
SGI is a program for treatment of statistical graphics. With this program we can
display dynamic graphics of multivariate data And, this enables users to visually
find trends, clusters, patterns, outliners, and other features of multivariate data
Using SGI, we can edit data files and display graphic plots. Source program is
constructed by MS-C language. Hardware environment is 386 machine with 640×480
resolution VGA monitor. Mouse is equipped for high user-interface interaction.
In SGI the user assigns any three variables from a data set to the three axes of the
display, producing a three-dimensional scatter plot of the data Three-dimensional
scatter plot of the data Three-dimensional graphics routines include real-time point
cloud rotation. The scaling icons expand or shrink data points during rotation.
Animation, commonly called slicing, on a 4th variable controlled by a sliding bar.
To utilize categorical information in the analysis the user can highlight certain
subset of the data Using a cursor-controlled selection box (brush), we can select
various points and change their color or shape. To utilize text information, the user
can identify points-select points in the display and have the corresponding names
displayed, and locate names-select names in a list and have the corresponding points
highlighted in the scatter plot window. Window dragging and window scaling are could
be done. SGI does not require leaning a language, memorizing any commands, or any
ability to type. Its windows user interface means that program operation requires
only pointing and clicking with a mouse.
3.1 Starting of SGI.
In MS-DOS operating system, execution SGI file shows fig 3-1. pointing mouse cursor
on 'Help' and clicking with a mouse, then small window appear to help user to
manipulate this software.
3.2 Make data files and data management.
Pointing mouse cursor on 'Edit' and clicking with a mouse, then you can edit data
files. Data formats are following. In first row, write variable numbers and maximum
data number. In second row write variable names and data numbers are following the
next rows. If terminate data editing, save the 'File' with the file name which has
the form 'file name.DAT' Pointing mouse cursor on 'Load' and clicking with a mouse,
then submenus are loaded in the left side. Pointing mose cursor on 'Load'and clicking
with a mouse, then data Files are loaded in the right side. And sellect data files
for statistical treatment. 21-data files could be loaded and which are restricted by
the available memory size.
3.3 Plot, Display and Draw
The functions of submenus are briefiy summarized.
*Statistic: Statistical routines such as T-test, Anova Regression etc are treated.
*Vert-Bar: Plots vertical histogram of one dimensional datas.
*Hor-Bar: Plots horizontal histogram of one dimensional datas.
*3D-Bar: Plots three-dimensional bar graph.
*Pie: Plots pie charts.
*Scattern: Plots N-dimensional scattern plots of N-dimensional data
*3D-Scatter: Plots 2-dimensional data in 3-dimensional space.
*Map: Plots with color and density such as data of population.
*3D-Map: Draw map on the sphere.
*Identify: Labelling and locating are manipulated.
*Linking: Visually link related points which are located in different scatter plots.
*Rotation: With rotation, data cloud's features are analyzed.
*Deleting: Delete outliers in graphic.
*Grouping: Proceeds cluster analysis by the method of scaling, coloring and rotation.
*Transform: Transform variables.
The results of the data analysis are ploted in the window. We add font, color, and
various lines, etc. This procedure make the result more precisely percepted and can
3.4 Feedback of Plot and Display
Plots and Display are played commutatively, and we can efficiently analyze
statistical data See figure 3-1.