Methods of Detecting Outliers in A Regression Analysis Model

This study detects outliers in a univariate and bivariate data by using both Rosner’s and Grubb’s test in a regression analysis model. The study shows how an observation that causes the least square point estimate of a Regression model to be substantially different from what it would be if the observation were removed from the data set. A Boilers data with dependent variable Y (man-Hour) and four independent variables X1 (Boiler Capacity), X2 (Design Pressure), X3 (Boiler Type), X4 (Drum Type) were used. The analysis of the Boilers data reviewed an unexpected group of Outliers. The results from the findings showed that an observation can be outlying with respect to its Y (dependent) value or X (independent) value or both values and yet influential to the data set.