Analisis regresi menggunakan metode

Analisis regresi menggunakan metode kuadrat terkecil (khususnya teknik Ordinary Least Square/OLS) sering memberikan hasil yang kurang tepat, dimana asumsi-asumsi klasik tidak terpenuhi, salah satunya yaitu tidak terjadinya multikolinieritas. Multikolinearitas muncul akibat adanya korelasi linier yang tinggi diantara dua atau lebih variabel bebas. Masalah yang tidak diinginkan ini dapat diselesaikan dengan menggunakan metode regresi gulud yang memberikan estimasi parameter bias. Regresi gulud memperkecil estimasi kuadrat terkecil untuk koefisien regresi menuju asal, memungkinkan dengan bias tetapi menyediakan varian yang lebih kecil. Namun, pilihan memperkecil k parameter dalam regresi gulud merupakan masalah serius lain. Algoritma baru berdasarkan Particle Swarm Optimization (PSO) diusulkan untuk menemukan optimal dari parameter yang diperkecil.

Results (English) 1:

Regression analysis using the smallest squares method (specifically the technique of Ordinary Least Square/OLS) often give results that are less precise, where classical assumptions are not met, one of them that is not the onset of multikolinieritas. Multicollinearity arises due to the linear correlation is high among two or more variables. This unwanted problem can be solved using the method of regression gulud that provide parameter estimation bias. Gulud minimize the quadratic regression estimation of regression coefficients for the smallest heading origin, allowing with bias but provides smaller variants. However, the option minimize the k parameters in the regression gulud other is a serious problem. A new algorithm based on Particle Swarm Optimization (PSO) proposed to find the optimal parameters of the scaled down.

Results (English) 2:

Regression analysis using the least squares method (especially engineering Ordinary Least Square / OLS) often provide less precise results, where classical assumptions are not met, one of which is the absence of multicollinearity. Multikolinearitas arising from high linear correlation between two or more independent variables. Unwanted problems can be solved by using regression method gulud are on parameter estimation bias. Regression gulud far the least squares estimation for the regression coefficient to the origin, allowing the bias but provides a smaller variant. However, the selection of parameter k in the regression far gulud is another serious problem. The new algorithm based on the Particle Swarm Optimization (PSO) is proposed to find the optimal parameters are minimized.