Descriptive statistics

Concepts

These components produce a statistical summary of the dataset.

Attributes status

None.

Descriptive statistics components

Component Description Parameters Note

Cross tabulation
Cross tabulation of attributes.

A. Agresti,"Categorical data analysis", John Wiley, 1990.

- Sort results from attribute name or statistical indicators.
- Attribute selection : (1) "target" = row, "input" = column, or (2) crossing of input attributes.
- Thresold for cells highlight.
- Discrete attributes only.

Linear correlation
Compute linear correlation of attributes.

S. Aïvazian, "Etudes statistiques des dépendances", Edition MIR, 1978.

- Sort results from attribute name or statistical indicators.
- Attribute selection : (1) "target" = row, "input" = column, or (2) crossing of input attributes.
- Continuous attributes only.

One-way ANOVA
One way analysis of variance.

- W.C. Guenther, "Analysis of Variance", Prentice-Hall, 1966.
- J.C. Bergonzini, "Analyse et planification des expériences", Masson, pp.43-51, 1995.

- "Target" : measurements, "input": factors.
- Sort results.
- Other descriptive statistics.
- "Target" must be continuous.
- "Input" must be discrete.

Univariate discrete stat
Descriptive statistics for discrete attributes.

- Select input attributes with "define status" component.
- Sort results.
- Input must be discrete.

Univariate continuous stat
Descriptive statistics for continuous attributes.

- Select input attributes with "define status" component.
- Sort results.
- Input must be continuous.

Non-parametrical test (KW)
KRUSKAL et WALLIS non-parametrical population comparison.

- S. Siegel, "Nonparametric statistics for the behavioral science", McGraw-Hill, 1956. - J.C. Bergonzini, "Analyse et planification des expériences", Masson, pp.62-63, 1995.

- Target : measurements, input : factors. - Target are continuous
- Input are discrete.

Group characterization
Group characterization, for instance, allow to describe group define by a clustering algorithm.

- L. Lebart, A. Morineau, M. Piron, "Statistique exploratoire multidimensionnelle", Ed. Dunod, pp. 181-184, 2000.

- With "define status": target is discrete, input can be continuous or discrete.

Last modification : January 21st, 2004.