The Mode Estimation Method of Inverse Gaussian Data

Abstract

The objectives of this research are to propose the mode estimation method of Inverse Gaussian data and to compare three mode estimation methods. Those methods are simple method, adjusted mode method with term (n+1)/n, and adjusted mode method with term (n+2)/n. The performance of each method can be measured by the absolute bias (|Bias|) and the mean square errors (MSE). This study was performed by using different sample sizes (n) of 5, 10, 20, 30, and 50 whereas the parameter  is fixed to be 1 and parameters  are fixed to be 1, 3, 5, 10, 15, and 20. This research used the Monte Carlo Simulation technique. The experiment was repeated 50,000 times for each condition. The results showed that the adjusted mode method with term (n+2)/n gave the lowest value of |Bias| and MSE in most of the criterions of sample sizes and parameter values.