[Forum SIS] Avviso di seminario: Lauro a DISMEQ (Dipartimento di Statististica e Metodi Quantitativi, Milano-Bicocca)
Fulvia Pennoni
fulvia.pennoni a unimib.it
Ven 16 Ott 2015 17:28:36 CEST
-----------------------------------------------------------------------------
Dipartimento di Statistica e Metodi Quantitativi
Via Bicocca degli Arcimboldi, 8 - 20126 Milano
-----------------------------------------------------------------------------
“Non symmetrical approach to component-based SEM”
Prof. Carlo Lauro
University of Naples “Federico II”, Naples, Italy
Mercoledì 21 Ottobre ore 14.30 Ed. U7 IV piano (stanza 4064)
__________________________________________________________
Abstract:
Structural Equation Modeling (SEM) is a multivariate analysis technique
that allows us to analyze relationships among several blocks of observed
variables.
PLS-PM has some inconsistencies in terms of coherence with the direction
of the relationships specified in the structural model (i.e., the
dependence relationships between LVs).
PLSPM algorithm analyzes relationships between blocks symmetrically and
misses to distinguish between dependent and explanatory blocks. As a
consequence, there is often a difference between what PLS-PM wants to
model and what is actually computed by the PLSPM algorithm.
We propose a new algorithm that takes into account explicitly the
directions of relationships in the structural model, based on the
maximization of the ex-plained variance of the MVs of the endogenous
blocks by the components of the explanatory blocks.
The new approach is more suitable for prediction purposes. As a measure
of the quality of the global model we propose a goodness of prediction
index based on redundancy criterion and prediction capability.
__________________________________________________________
Si allega la locandina.
Cordiali saluti
Fulvia Pennoni
-------------- parte successiva --------------
Un allegato non testuale è stato rimosso....
Nome: Lauro.pdf
Tipo: application/pdf
Dimensione: 76234 bytes
Descrizione: non disponibile
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20151016/a80573cc/attachment-0001.pdf>
Maggiori informazioni sulla lista
Sis