Browsing by Subject "Finite mixture models"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item A finite mixture approach for household residential choices(2010-12) Shiroya, Michael; Farmer, Michael; Belasco, Eric J.; Elam, Emmitt; Chidmi, BenaissaIn the housing sector, on the demand side, attempts to characterize the nature of housing demand have been primarily implemented through hedonics whereby hedonic price functions relate market value to residential housing stock. Hedonic price models, however, do not identity sub-markets and thus may represent the workings of housing market and the valuation of amenities and dismaenties in a housing market. Identification of sub-markets allows us to have a more precise and reliable understanding of housing markets in general The challenge then becomes how to identify market segments and impute the marginal value of housing characteristics in different market segments. In this paper, I will attempt to identify and delineate residential sub-markets in an array of residential stock demand and derive the predicted marginal value of attributes of housing stock within all sub-markets. These sub-markets can be thought of composing a homogenous ‘type’ of households. Typing households in a statistically is a useful way of accounting for preference and utility differences in a structurally sound manner that offers deeper insight into the welfare of consumers of residential stock. I accomplish the typing of households by implementing a finite mixture model which gives a probability distribution of an individual household being a particular type. The model best fitted the array of households into two types. The types exhibited differences in their attitudinal and demographic characteristics.Item A new approach of genetic-based EM algorithm for mixture models(2011-05) Abeysundara, Sachith P.; Seo, Byungtae; Trindade, A. AlexandreFinite mixture models have been receiving an important attention over the years in practical and theoretical point of view, but it is still challenging task to estimate a reasonable estimator based on the maximum likelihood method. The most widely used technique to solve the problem, to some extent, is the EM algorithm. Researchers have done a lot of work to improve the results of the EM algorithm by modifying its basic idea. This work present such an attempt to obtain better estimates for a finite normal mixture model. A traditional evolutionary technique, known as Genetic algorithm, is coupled with the EM algorithm to improve the estimates of the EM algorithm started with a random initial vector of parameters. The presented method is tested with the availability of a Non-penalized and Penalized likelihood functions. Based on results, we can see that the proposed method is always superior to the classical EM algorithm when one concerns the global maximizer in the mixture likelihood function.