Analysis of Classic assumption test and multiple linear regression coefficient test for employee structural office recommendation

摘要

Religious High Court Bandar Lampung is a Religious Courts Institution in charge of carrying out the settlement of cases at the appeal level. Promotion is one of the career development through employee performance appraisal. Currently, the performance appraisal process has not been carried out objectively, there are employees who occupy structural positions but do not meet the competencies and promotion principles that still prioritize the like and dislike element so that it has an impact on giving promotions to positions within the judiciary. Utilization of data mining using the Multiple Linear Regression method provides predictions of recommendations for employees who are entitled to occupy positions in the agency. Multiple Linear Regression Method is a regression technique that produces equations and will explain the relationship patterns of the variables used. Implementation of the use of the Multiple Linear Regression method obtained from 40 employees there are 26 employees who deserve to be given a promotion recommendation. The results of the regression performance test using 10-cross validation with the WEKA tools that present the error value of the prediction results using the correlation coefficient value (Correlation Coefficient) is 0.8066 or 80.66% resulting in an MAE Error (Mean Absolute Error) value of 2.24% and RMSE (Root Mean Square). Error) 3.88%. These results explain that the linear regression method has a good performance with a small error rate in the regression performance evaluation test. The results of the classical assumption test analysis concluded that the data were normally distributed and the regression could be met properly by performing multicoinierity tests, heteroscedasticity tests and normality tests. As for the analysis of the multiple linear regression coefficient test which was carried out with the T test and the F test, it resulted in a test that had a significant effect on employee performance variables so that this research was feasible to continue.