Mencari Nilai Statistik z, t, Chi Square, dan f secara Heuristik Berdasarkan Dynamic Significance Level

Abstract

The value of statistics z, t, χ2, and f are widely used in estimation problem, especially in construction of confidence interval with regard to in what interval that the parameter lies, and it’s used in hypothesis testing as well. In order to conduct these method, one need the critical value of zα, tα, χ2 α, and fα at level of significance α, and these critical values cannot be computed exactly since its (measure) function P−1 : R → Ω does not exist. Nevertheless, these critical value can be searched heuristically for dynamic significance level α. In this research, the heuristic procedur is designed based on binary search, bisection method, and tabu search in order to find these critical statistic’s values. The searching of these critical values start from given any interval that these values lies with, and then use oscillation strategy, and redeem it by declare a tabu not any interval of random number that’s used, repeat it untill the value of these critical value of statistics is found respect with given tolerance quantity ϵ, or termination criteria is meet. After that, interpolate the statistic’s values by taking the mid point of the interval as a return value of this heuristics procedur.