Outlier detection in multivariate data

The objective of this research is detection of outliers in multivariate data employing various distance measure, particularly using robust regression diagnosis technique. Several classical outlier identification methods are based on the sample mean and covariance matrix in general. But they do not always yield better result, as they themselves are affected by the outliers. Sometimes one outlier point has hide the other outliers. To identify them, methods which have masking effect with outlier points are being used. An appropriate method is adopted to identify the unmasking outliers and also to compare the various distance measures. Mathematics Subject Classification: 62H99, 62J05