DISCOVERING PATTERNS AND DEVIATIONS IN DATA: COMPARISON OF ANOMALY DETECTION PROCEDURE IN REGRESSION

Finding interesting or unusual patterns plays an important role in a lot of fields, including regression, business intelligence, data mining, and decision making. In many domain applications, anomaly detection and regression are both important, with the aim of identifying patterns or trends that can pose challenges in their respective areas. . As a result, empirical analysis was done to find out how well leverage, DFFITS, and standardized residual performed in identifying abnormalities. Findings. Mostly, for discovering patterns of data in cases regarding simple and multiple linear regression, if the data is flagged as high leverage, then there is a possibility that it may highly influence the estimated regression. However, in both the cases, there are hardly any side effects from the performance of R . It emphasizes that there 2 is still a strong relationship between independent and dependent variables. Besides, if the data is flagged as influential or has a high standardized residual, it is found that the model adequacy has increased R 2 = 0.4324 to R 2 = 0.6877 and R 2 = 0.7056 to R 2 = 0.7532.