Automated Noise Detection in a Database Based on a Combined Method

Data quality has diverse dimensions, from which accuracy is the most important one. cleaning one of preprocessing steps in data mining consists detecting errors and repairing them. Noise a common type error, that occur database. This paper proposes an automated method based on k-means clustering for noise detection. At first, each attribute (Aj) temporarily removed applied to other attributes. Thereafter, k-nearest neighbors used cluster. After value predicted Aj record by nearest neighbors. The proposed detects noisy attributes using values. Our able identify several noises record. In addition, this can detect fields with different types, too. Experiments show averagely 92% existing data. compared detection association rules. results indicate have improved 13%.