For simplifying the fuzzy-weighted quasi-likelihood estimation procedure and reducing the influence of outliers in quasi-likelihood estimation, optimal fuzzy clustering analysis method and dimension reduction technique are combined to de- velop a modified fuzzy -weighted quasi-likelihood estimation method. Under some weak conditions, the asymptotic properties of modified fuzzy-weighted quasi-likelihood estimator are derived. Simulation results show the modified fuzzy-weighted quasi-likelihood estimator performs well in finite sample. Moreover, the method is studied on several examples.
展开▼
