Rules in incomplete information systems
Information Sciences: an International Journal
Rough Sets: Theoretical Aspects of Reasoning about Data
Rough Sets: Theoretical Aspects of Reasoning about Data
A Comparison of Several Approaches to Missing Attribute Values in Data Mining
RSCTC '00 Revised Papers from the Second International Conference on Rough Sets and Current Trends in Computing
On the Extension of Rough Sets under Incomplete Information
RSFDGrC '99 Proceedings of the 7th International Workshop on New Directions in Rough Sets, Data Mining, and Granular-Soft Computing
On the Unknown Attribute Values in Learning from Examples
ISMIS '91 Proceedings of the 6th International Symposium on Methodologies for Intelligent Systems
Local and global approximations for incomplete data
Transactions on rough sets VIII
A Local Version of the MLEM2 Algorithm for Rule Induction
Fundamenta Informaticae - Understanding Computers' Intelligence Celebrating the 100th Volume of Fundamenta Informaticae in Honour of Helena Rasiowa
Local and global approximations for incomplete data
RSCTC'06 Proceedings of the 5th international conference on Rough Sets and Current Trends in Computing
| Hi-index | 0.00 |
This paper presents results of experiments on incomplete data sets obtained by random replacement of attribute values with symbols of missing attribute values. Rule sets were induced from such data using two different types of lower and upper approximation: local and global, and two different interpretations of missing attribute values: lost values and "do not care" conditions. Additionally, we used a probabilistic option, one of the most successful traditional methods to handle missing attribute values. In our experiments we recorded the total error rate, a result of ten-fold cross validation. Using the Wicoxon matched-pairs signed ranks test (5% level of significance for two-tailed test) we observed that for missing attribute values interpreted as "do not care" conditions, the global type of approximations is worse than the local type and that the probabilistic option is worse than the local type.