An Algorithm for Detecting Noise on Supervised
Luis Daza, Edgar Acuña · 2007
In this paper, we introduce, a new algorithm, QcleanNoise, for detecting noisy instances. The effects of the algorithm in three supervised classifiers: LDA, KNN and RPART, a decision tree classifier, are discussed. Comparison with other procedures is carry out on four well known Machine Learning datasets. The experimental results shows that our algorithm performs better than current procedures detecting efficiently noisy instances using less computational time.