Robust regression and outlier detection
Robust regression and outlier detection
Hi-index | 0.00 |
Outliers are of concerned by many researchers since they could possibly provide very useful information or obscure the final time series model. Robust methods are often used to reduce the effect of outlying observation in the model. In this study, we identify the type of outliers using classical least squares method and robust least median square method for 104 weekly mean water level of Langat River from January 2002 to December 2003. The results based on the two techniques are compared and it is found that the robust least median squares method successfully unmasked the effect of outlier as compared to the least squares method.