Estimating the general chain logit model using parametric and nonparametric methods using simulation

Abstract

This research includes estimating the distribution of theGeneralized logarithm series the two-parameter parameters, one of the distributions Discrete has three modalities adopted in the estimate is method Maximum likelihood and the method of moments and the method proposed include the use of search Cuckoo algorithm techniques to generate test data automatically and through the estimation parameters specimen user where and conducted simulation experiments when volumes of samples (n = 100,250,400) and repeated each experiment (L = 1000) and the use of average squares error (MSE) and the Absolute mean percentage error (MPE) to compare the estimators have presented the results of the estimation of the two parameters (α, β).