The most important Factors affecting the accident, according to the standard Bayesian future | ||
| journal of kerbala university | ||
| Article 1, Volume 8, Issue 1, May 2012, Pages 127-134 PDF (0 K) | ||
| Abstract | ||
| It is known there are many statistical criteria that will help in the selection of the best models and which ones can choose the model which includes the smallest number of explanatory variables to obtain the same information (roughly) that you get if it included the model of all explanatory variables, as well as the model chosen by these standards characterized by many desirable statistical properties. One of the main criteria used to select models is Normalized Maximum Likelihood (NML ) and Criterion Bayesian future(BFC(Resulting from the integration of normative: Akaike’s Information Criterion (AIC )& Bayesian Information Criterion (BIC) to be a more accurate measure. In addition to the expense of the coefficient of determination( R Square) and compared with the average coefficient of determination R2 Adjusted of the partial groups in linear regression models with the coefficient determining R2 Cox, Snell and determine the coefficient R2 Nagelkerke in models of logistic regression and which represent the proportion of variance unexplained variables independent as each model is taken Malay despite differences in the interpretation of variation in each in the case of linear regression the traditional logistic regression. To analyze data pertaining to the phenomenon of accidents and to study factors affecting them by identifying the impact of all variables on the logistic model and the exclusion of important variables in the model study to be reduced to the lowest possible number and composition of a subset of them give information sufficient illustrative of the phenomenon studied.Adopted as the sample (135) of patients coming to emergency Hussein Hospital in Karbala, the holy year, and the information was collected through statistical form included a number of questions to measure a set of variables that affect the fate of an infected person. Was the reason of the person at the scene (business, entertainment, a visit), the cause of the accident (traffic accident, fall from a height, firearms, fire, sharp object, sinking, other), gender, type of injury (bruise, wound, fracture, internal bleeding), the extent of injury (sharp, simple), the place of infection. In order to determine which variables are more important than others primarily affect the fate of the person that came out an improved or enter the lobby or Log in intensive care Accordingly, the research found that the (fate of the person) is linked (type of injury, the extent of injury) more than others and match the performance of the Criterion Bayesian future(BFC(in the selection of models with the Normalized Maximum Likelihood and that the coefficient R2 Nagelkerke is the best in the interpretation of the model and its value is equal to 79% according logistic model. | ||
|
Statistics Article View: 95 PDF Download: 54 |
||