University of Konstanz, Center of Finance and Econometrics (CoFE), Konstanz
A three parameter (location, scale, shape) generalization of the logistic distribution is fitted to data. Local maximum likelihood estimators of the parameters are derived. Although the likelihood function is unbounded, the likelihood equations have a consistent root. ML-estimation combined with the ECM algorithm allows the distribution to be easily fitted to data.
ECM algorithm generalized logistic distribution location-scale-shape model maximum likelihood estimation