Artificial Intelligence
Robust Learning with Missing Data
Machine Learning
An Electromagnetism-like Mechanism for Global Optimization
Journal of Global Optimization
Incremental Induction of Decision Trees
Machine Learning
Machine Learning
Nearest neighbour approach in the least-squares data imputation algorithms
Information Sciences: an International Journal
On Classification with Incomplete Data
IEEE Transactions on Pattern Analysis and Machine Intelligence
Principal component analysis for data containing outliers and missing elements
Computational Statistics & Data Analysis
An Algorithm for Classifying Incomplete Data with Selective Bayes Classifiers
CISW '07 Proceedings of the 2007 International Conference on Computational Intelligence and Security Workshops
A selective Bayes Classifier for classifying incomplete data based on gain ratio
Knowledge-Based Systems
Mining Predictive k-CNF Expressions
IEEE Transactions on Knowledge and Data Engineering
Applying electromagnetism-like mechanism for feature selection
Information Sciences: an International Journal
Hi-index | 0.01 |
Numerous classification approaches have been proposed; however, most of them address complete data problems. Because actual data sets are typically incomplete for various reasons, algorithms for classification with incomplete data have received increasing attention and numerous methods have been developed to address incomplete data. These approaches have certain drawbacks or the pre-assumption of data missing at random, which is difficult to verify. Ramoni and Sebastiani presented the Robust Bayes Classifier (RBC) to eliminate the assumption. Nevertheless, RBC assumes that the attributes are independent for each class. A broken assumption degenerates classification performance. Therefore, to find the feature subset with the best performance is the top priority. Because selecting features belongs to NP-complete problems, this study combined Electromagnetism-like Mechanism algorithm with RBC for feature selection and classification tasks with incomplete data. A numerical experiment on 11 incomplete data sets was conducted. The results indicated greatly improved RBC performance combined with each feature selection approach. The proposed hybrid method outperformed the other algorithms not only in balanced classification accuracy, but also in efficiency of feature selection.