Abstract:
The purpose of this research is to compare the parameter estimator in polynomial regression models by Ordinary least squares method (OLS), ridge ordinary least squares method (ROLS) and ridge least absolute value method (RLAV). The criterion of comparison is average mean squares error (AMSE) and use ratio of different average mean squares error (DIFF) to compare the efficience of these methods. The distribution of random errors are normal distribution with mean equal to 0 and variance equal to 4, 6, 8 and 10, respectively. The sample sizes used in this study are 15, 30, 60, 120 and 240, respectively, highest degree of independent variables for dependent variable building in model (MB) are 2, 3, 4, 5 and 6, respectively. The data for this experiment are generated through the Monte Carlo simulation technique and repeating 1,000 times for each case. The results of this research are as follow : The average mean squares error (AMSE) vary with, most to least, respectively, highest degree of independent variables for dependent variable building in model (MB), variance of random errors (σ²) but AMSE is converse to sample size (n). 1. In case of sample size is low (n 30). For all MB and all variance of random errors (0%) the RLAV method is the best, the ROLS method and the OLS method are secondary method respectively. 2. In case of sample size is medium (n=60). If MB is low (MB ≤3). variance of random errors is low (σ² <6), the lowest AMSE is the RLAV method and the highest AMSE is the OLS method but the DIFF is less than one times. Therefore OLS method is suitable. For the other case, the RLAV method is the best, the ROLS method and the OLS method, respectively. 3. In case of sample size is quite more (n = 120). If MB is low (MB ≤3), for all variance of random errors (σ²), and If MB = 4, variance of random errors is low (σ²E<6), the lowest AMSE is the RLAV method and the highest AMSE is the OLS method but the DIFF is less than one times. Therefore OLS method is suitable. For the other case, the RLAV method is the best, the ROLS method and the OLS method, respectively. 4.In case of sample size is high (n = 240). If MB is low to medium , for all variance of random errors, the lowest AMSE is the RLAV method and the highest AMSE is the OLS method but the DIFF is less than one times. Therefore OLS method is suitable. If MB is so high (MB ≥5), for all variance of random errors (σ²) the RLAV method is the best. the ROLS method and the OLS method, respectively.