Page 304 -

Appendix A. CD Datasets 293

PRM : Average perimeter of the defects (in pixels)=PRT/N
ARM : Average area of the defects (in pixels)=ART/N
NG : Number of big defects (area bigger than an adequate threshold)
PRTG: Total perimeter of big defects (in pixels)
ARTG: Total area of big defects (in pixels)
RAAR: Area ratio of the defects = ARTGIART
RAN : Ratio of the number of defects = NG/N
Source: A Campilho, Engineering Faculty, Oporto University.
Reference: A Campilho (1985), Ph.D. Thesis, Engineering FacultyIDEEC, Oporto
University.

A.4 Crimes

The Crimes.xls file contains the number of crimes per thousand inhabitants in
Portuguese districts during 1988, categorized as:

CAPeople - Crimes against people
CAProperty - Crimes against property

Source: "DiBrio de Noticias" collection, 1989.

A.5 Cardiotocographic Data

The CTG.xls file contains measurements and classification results of
cardiotocographic (CTG) examinations of 2126 foetuses. The examinations were
performed at S. Jog0 Hospital, Oporto. Cardiotocography is a popular diagnostic
method in Obstetrics, consisting of the analysis and interpretation of the following
signals: foetal heart rate; uterine contractions; foetal movements.
The measurements included in the CTG.xls file correspond only to foetal heart
rate features (e.g., basal value, accelerativeldecelerative events), computed by an
automatic system. The classification corresponds to a diagnostic category assigned
by expert obstetricians independent of the CTG.
The following cardiotocographic features are available in the CTG.xls file:

LBE baseline value (medical expert) LB baseline value (system)
AC accelerations FM foetal movement
UC uterine contractions
ASTV percentage of time with abnormal short term (beat-to-beat) variability
MSTV mean value of short term variability
ALTV percentage of time with abnormal long term (one minute) variability
MLTV mean value of long term variability
DL light decelerations DS severe decelerations