การเปรียบเทียบตัวประมาณค่าพารามิเตอร์แสดงตำแหน่งและสเกล ของการแจกแจงแบบปกติที่มีค่าผิดปกติ / ดารณี ตั้งโชฏิกะ = A comparison on estimators Of location and scale parameter Of normal distribution having outliers / Daranee Thangchotika
The objective of this research is to compare the estimators of location and scale parameter of normal distribution having outliers. In this study, the estimators are robust estimator, bootstrap estimator, and unbiased estimator. They were compared by mean square error (MSE). The comparison was done under 2 levels of data having outliers: data have mild outliers, and data have extreme outliers. The sample sizes used in this research are 20, 30, 50, and 70. The proportion of contamination are 0.05,0.10, 0.15, and 0.20. The research was done through the Monte Carlo technique repeating 500 times for each situation. The conclusions of this research are as follows: 1. Data have mild outliers. In the case of estimating a location parameter, when data have location contamination in normal distribution, the robust estimator and the unbiased estimator have nearly the same MSE, but when data have scale contamination in normal distribution , the robust estimator and the bootstrap estimator have nearly the same MSE. For estimating a scale parameter, the robust estimator has the lowest MSE in most situations. 2. Data have extreme outliers. In the case of estimating a location parameter, the robust estimator has the lowest MSE in all situations under study. For estimating a scale parameter, the robust estimator has the lowest MSE in all situations under study.