1,720,998 research outputs found
BAYESIAN NETWORKS FOR FINITE POPULATIONS
No full textIn questo lavoro si propone uno stimatore della distribuzione di frequenze congiunta quando il campione è estratto con disegni campionari complessi tramite le reti Bayesiane. Questa formalizzazione può essere utile ai fini della valutazione delle politiche da parte di un decision maker. Un esempio nel settore dell'agricoltura viene discusso.In this work an estimator of the joint frequency distribution is defined when the sample is drawn according to complex survey designs by means of the Bayesian networks. This formalization can be useful for the evaluation of policies by a decion maker. An example in the agricultural sector is discussed
Estimation of contingency tables in complex survey sampling using probabilistic expert systems
No full textIn this paper we explore the possibility to use a particular class of models, known as
probabilistic expert systems, to define two classes of estimators of a contingency table in case of stratified sampling designs. The two classes are characterized by the different role of the sampling design: in the first, the sampling design is treated as an additional
variable; in the second, it is used only for estimation purposes by means of the survey weights. The bias/variance tradeoff of the estimators is analyzed and the
consequences of model misspecification are illustrated. Furthermore,it is shown that the Horvitz–Thompson estimator belongs to both classes of estimators. It comes out that theHorvitz–Thompson estimator is almost always inefficient but robust. Monte Carlo simulations illustrate the efficiency of the proposed estimators
COHERENCE OF SAMPLE ESTIMATES FOR FINITE POPULATIONS: SOME RESULTS BASED ON BAYESIAN NETWORKS
No full textThe idea of coherence of estimates is crucial in the production process of the national
institutes of statistics. In this paper we consider the idea of internal and external coherence. The first
concept is related to results of a single survey while the second is related to results of different surveys.
It will be shown how Bayesian networks can be usefully exploited in order to define general estimators
in a finite population context
Model assisted approaches to complex survey sampling from finite populations using bayesian networks
No full textA class of estimators based on the dependency structure of a multivariate variable of
interest and the survey design is defined. The dependency structure is the one described by the Bayesian networks. This class allows ratio type estimators as a subclass identified by a particular
dependency structure. It will be shown by a Monte Carlo simulation how the adoption of the estimator corresponding to the population structure is more efficient than the others. It will also
be underlined how this class adapts to the problem of integration of information from two surveys
through probability updating system of the Bayesian networks
Model Assisted Approaches to Complex Survey Sampling from Finite Populations Using Bayesian Networks: a Tool for Integration of different Sources
No full textA class of estimators based on the dependency structure of a multivariate variable of interest and the survey design is
defined. The dependency structure is the one described by the Bayesian networks. This class allows ratio type estimators as
a subclass identified by a particular dependency structure. It will be shown by a MonteCarlo simulation how the adoption
of the estimator corresponding to the population structure is more efficient than the others. It will also be underlined how
this class adapts to the problem of integration of information from two surveys through the probability updating system of
the Bayesian networks
Propagazione dell’informazione nel campionamento da popolazioni finite: reti bayesiane e poststratificazione
No full textOne characteristic of Bayesian networks is the possibility to apply efficient
algorithms for updating the estimates of a probability distribution when auxiliary
information is available. In this paper the poststratification estimators of a joint
distribution function are represented with the help of Bayesian networks
Some Results on Probabilistic Expert Systems for Finite Population Sampling
No full textLa scelta della strategia campionaria disegno/stimatore privilegia l’uso
di variabili ausiliare connesse con le variabili di interesse. Le variabili ausiliare sono
utilizzate per definire un nuovo sistema di pesi campionari e/o uno stimatore opportuno.
Ballin et al. (2005) mostrano come gli usuali stimatori implicitamente fanno riferimento
a una struttura di dipendenza particolare fra variabili di interesse e ausiliarie: il modello
saturato. In questo lavoro si discutono alcune propriet`a degli stimatori suggeriti da
strutture di dipendenza fra variabili rappresentate mediante sistemi esperti probabilistici
Bayesian networks and complex survey sampling from finite populations
No full textWe propose a novel methodology based on the concept of Bayesian network (BN, see Cowell et al., 1999) for the estimation of
a joint probability distribution of a set of categorical variables when samples are drawn according to complex survey designs.
Note that, restricting ourselves to categorical variables, the previous aim corresponds to estimation of a contingency table, a
very frequent problem in Official Statistics.
BNs are graphical devices largely used in many different scientific contexts, such as artificial intelligence and multivariate
statistics (Neapolitan, 2004). However, when estimating and using BNs, observations have always been considered as i.i.d.
generations from a suitable joint distribution function. Up to now, BNs have never been defined and applied when sampling
from finite populations.
This paper shows that BNs can be easily adapted to the context of finite survey sampling via the definition of a suitable additional
variable, in the following denoted with SD, representing the survey design. Hence, SD will be a categorical variable with as
many states as the different inclusion probabilities of first order. The BN representation allows the definition of a much larger
class of estimators, of the model assisted type (see Sarndal ̈ et al., 1992). Also, the possibility to use poststratification methods
and, in general, integration of different surveys is illustrated
- …
