Graphics Reference
In-Depth Information
modeling without exploring the data first is the purist one of satisfying the ideal-
ist prerequisite for a hypothesis test - although whether that is relevant for any real
analysis is moot.
Graphics are essential for exploratory work. hey provide overviews and insights
that complement statistical summaries. On the other hand, graphical analyses with-
out follow-up analytic support remain inconclusive. Visualization results tend to be
qualitative rather than quantitative, general rather than precise. Statistical model-
ing can assess the strength of evidence supporting ideas generated from graphical
EDA and help to define those ideas more exactly. On top of that, statistical modeling
can tease out more complex relationships that are not immediately visually apparent.
However, there is not much point in looking for complex relationships if the quality
of the data is in doubt, which is one of the reasons that modeling benefits from prior
data visualization. Modeling also benefits from graphical support ater analysis, in
relation to investigating residual patterns for individual models and comparing and
combining groupsofmodels.hesearediscussedinotherchaptersintheHandbook.
Software
3.10
hesotware used for mostof thedisplaysinthis paper wasMartinheus'sMondrian
(http://stats.math.uni-augsburg.de/Mondrian/). Other sotware was used at various
stages to assist with data cleaning and restructuring.
Acknowledgement.
Financial support from the Deutsche Forschungsgemeinschatvia SFB
“Ökonomisches Risiko” is gratefully acknowledged. hanks also to Rouslan Moro, Doro-
thea Schäfer and Uwe Ziegenhagen for helpful comments on earlier drats.
References
Hofmann, H. (
). Exploring categorical data: interactive mosaic plots, Metrika,
(
):
-
.
Hofmann, H. and heus, M. (submitted). Interactive graphics for visualizing condi-
tional densities, Journal of Computational and Graphical Statistics.
Inselberg, A. (
). Don't panic ... do it in parallel, Computational Statistics,
(
):
-
.
Playfair,W.(
).Playfair's Commercial and Political Atlas and Statistical Breviary,
Cambridge, London.
Tukey, J. (
). Exploratory Data Analysis, Addison-Wesley, London.
Unwin,A.(
).Requirementsforinteractivegraphicssotwareforexploratorydata
analysis, Computational Statistics,
:
-
.
Unwin, A. R., heus, M. and Hofmann, H. (
). Graphics of Large Datasets,
Springer, Berlin.
