Database Reference
In-Depth Information
(a)
(b)
Fig. 2.
Some Results from Project One
The Student Work
The project was undertaken by a group of two students in 2010. Details of their work
are listed as follows:
•
Data understanding. Besides known facts about the attribute domains and the data
set size, virtually nothing more on data understanding was done. Some anomaly
records (farmers of age 3 and car insurance buyers under the legal minimum age
for driving) were spotted. No discovery objectives were mentioned.
•
Data preparation and pre-processing. Regional codes were replaced with nominal
labels (A, B, C, D). The unknown regional code 999 was converted to the un-
known symbol recognisable by Weka. The anomaly records were removed from
the data set. The justification given was that 21 anomalies count only 0.14% of
the total number of records. The values for age attribute were discretized using an
