Graphics Reference
In-Depth Information
in order to have different views of the same set of data, the choice of particular
range of values for a numerical variable, the use of tools to obtain a better visual
perceptionofthe data (inthecaseofmultiple points inthe same plotting location
one simple solution is jittering, i.e., a small amount of uniform random noise is
added to the data before graphing);
Asking for a representation of part of the information, namely, starting from unit
attributes (subsetselection) orquerying agiven region (selection of an area), that
is, the user can concentrate on particular points and regions. hese are made to
stand out from the background and highlighted using a different color, dimen-
sion, or filling in order to look for clusters or behavior different with respect to
unselected points;
Analyzing several linked views ofthe same data (linking, brushing,slicing).Mul-
tiple graphics is an old technique used to propagate information through differ-
ent plots that display different aspects (dimensions) of the same dataset. When
points are selected on a view through an input device, all views of that case are
highlightedoneachoftheotherplotssimultaneously. Inthiswaytheusercanan-
alyze if the selected points have some particular features in the other dimensions
or views. Linking, brushing, and slicing are based on the idea that by emphasiz-
ing the commonidentity of cases in multiple displays, the analyst can moreeasily
relate several displays to one another. hey allow one to look for clusters, to ex-
plore the relationships among variables, or to investigate if a particular pattern
or position is confirmed on different views of the same data. Moreover, the link-
ing capability can be used not only among different graphical displays but also
between a graph and the input data;
Redrawing the views using only certain selected levels of a strata variable (con-
ditional selection), mainly categorical variables, even if intervals of a quantitative
variable can be used. he redrawing of the graphs requires one to rerun the sta-
tistical analysis in order to visually concentrate on groups of units characterized
by the same level(s) of the strata variable. his tool provides a powerful method
to execute conditional analysis.
An Application: the Survey
of Italian Household Income and Wealth
4.8
Previous ideas are best transmitted through a typical application. Results shown in
thepresentsection havebeenobtained usingthesotware VITAMIN-S.VITAMIN-S
(VIsual daTA MINing System) is a prototype sotware realized in the framework
of the
-
IST-
-
project; the project was sponsored by EC
fith FP.
hiscontribution hasbeengreatly affectedbytheauthors' experienceinthedevel-
opment of the VITAMIN-S sotware. he most innovative aspect of VITAMIN-S
