The forward search is a general method to detect multiple outliers and to determine their effect on inference about models fitted to data. From the monitoring of a series of statistics based on subsets of data of increasing size we obtain multiple views of any hidden structure. Sometimes, some features emerge unexpectedly during the progression of the forward search only when a specific combination of forward plots is inspected at the same time. These features have to be harmonized and linked together in order to give an exhaustive description of a complex problem. In this paper, we use a set of new robust graphical tools on a mixture of logit regressions. We use simulated data and we show the dynamic interaction with different “robust plots” to highlight the presence of groups of outliers and regression mixtures in the context of logit regression and highlight the effect that these hidden groups provide on the fitted model.
La Forward Search `e un approccio di analisi innovativo robusto che parte da un insieme di dati ridotto, privo di unit´a anomale ed include sequenzialmente le rimanenti osservazioni in base ad una misura via via crescente di “anomalia” delle stesse. Attraverso una combinazione efficace di modellazione statistica e di grafici diagnostici, la forward search costituisce uno strumento potente che individua la presenza di valori anomali (singoli o raggruppati), e permette di valutare il loro effetto sui risultati delle analisi tradizionali. In questo studio proponiamo l’uso di tale approccio nella stima di un modello di regressione logistica dove dati simulati provengono da popolazioni binomiali differenti. Grazie all’analisi grafica sar´a pos- sibile identificare unit´a o gruppi anomali che solo apparentemente appartengono alla stessa popolazione.
Mixtures of Logit Regressions Detection with Forward Search
Bini M;Velucchi M
2014-01-01
Abstract
The forward search is a general method to detect multiple outliers and to determine their effect on inference about models fitted to data. From the monitoring of a series of statistics based on subsets of data of increasing size we obtain multiple views of any hidden structure. Sometimes, some features emerge unexpectedly during the progression of the forward search only when a specific combination of forward plots is inspected at the same time. These features have to be harmonized and linked together in order to give an exhaustive description of a complex problem. In this paper, we use a set of new robust graphical tools on a mixture of logit regressions. We use simulated data and we show the dynamic interaction with different “robust plots” to highlight the presence of groups of outliers and regression mixtures in the context of logit regression and highlight the effect that these hidden groups provide on the fitted model.File | Dimensione | Formato | |
---|---|---|---|
BDV.pdf
non disponibili
Dimensione
225.57 kB
Formato
Adobe PDF
|
225.57 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.