This dataset consists of 1930 instances and 50 attributes. Each instance corresponds to a graduate student. The first attribute is the sex of the student and the second one is the age. Each of the other 48 attributes is a graded degree of importance of different properties of the future job, evaluated by the student on an ordinal scale with 5 levels ranging from 1 (completely unimportant) to 5 (very important). Examples of such properties include "reputation", "safety", "high income", "friendly colleagues". Thus, every student was asked how important he or she considers these properties to be, and the student answered by assigning one of the aforementioned 5 levels. For privacy concerns, the names of those properties are not given in the data. If you have any problem regarding this data set, please do not hesitate to contact Weiwei Cheng.
Download (in .zip-format)
BeLa-E data have been used in:
- Weiwei Cheng, Krzysztof Dembczyński, Eyke Hüllermeier
Graded multi-label classification: the ordinal case [pdf][bibtex][slides][poster]
Proceedings of the 27th International Conference on Machine Learning (ICML-10): 223-230, Omnipress
Haifa, Israel, June 2010