ANALISIS ROBUST DARI MODEL LINEAR CAMPURAN TERGENERALISIR

wahyuning ning widiyastuti

Abstract


We will develops a technique for finding robust maximum likelihood estimates of the model parameters in generalized linear mixed models. The asymptotic properties of the robust estimates lies in the presence of outliers, and these estimates also compared to the ordinary classical estimates. By starting from a natural class of robust estimators for generalized linear models based on the notion of likelihood, we define robust deviances that can be used for stepwise model selection as in the classical framework. The binomial models are treated in detail. Application to real data and a sensitivity analysis show that the inference obtained by means of the new techniques is more reliable than that obtained by classical estimation and testing procedures.


Full Text:

PDF

Refbacks

  • There are currently no refbacks.


Iqtishadia Journal Indexed by :

http://journal.stainkudus.ac.id/indexing/crossref.jpghttp://journal.stainkudus.ac.id/indexing/moraref.jpghttp://journal.stainkudus.ac.id/indexing/ipi.jpghttp://journal.stainkudus.ac.id/indexing/academiaedu.jpghttp://journal.stainkudus.ac.id/indexing/googlescholar.jpghttp://journal.stainkudus.ac.id/indexing/sinta.jpghttp://journal.stainkudus.ac.id/indexing/issuu.jpghttp://journal.stainkudus.ac.id/indexing/cosmos.jpghttp://journal.stainkudus.ac.id/indexing/isjd.jpghttp://journal.stainkudus.ac.id/indexing/journalfactor.jpghttp://journal.stainkudus.ac.id/indexing/jifactor.jpghttp://journal.stainkudus.ac.id/indexing/journaltocs.jpghttp://journal.stainkudus.ac.id/indexing/scholarsteer.jpghttp://journal.stainkudus.ac.id/indexing/academiakeys.jpg