Using the greatest possible method in estimating the parameters of the partial hierarchical Poisson model (various inclinations) with practical application.

Abstract

In this research an analytical study on a multi-level model, a model of the partial hierarchical
Poisson regression (a model of different tendency) where this model is one of the most widely
applied models in the analysis of data characterized by the fact that the observations that take
a hierarchical form.
This study aims to study the factors that affect the phenomenon of increasing mortality due to
traffic accidents. The parameters of this model were estimated using FML method. The results of the maximum potential method were compared for Poisson partial hierarchy regression model
(different inclination). It included the practical application of data on deaths due to traffic
accidents in the governorates of Iraq and over two years from 2014 to 2015 issued by the
Central Statistical Organization of the Ministry of Planning in coordination with the Ministry of
Interior, where five stations were selected in Iraq h The first station represents the province of
Baghdad, the second station represents the province of Basra, the third station represents the
province of Kirkuk and the fourth station represents Dhi Qar The fifth station represents the
remaining provinces, where each province or group (the number of units of the second level)
random effect as each province represents Group where the first level represents deaths due to
traffic accidents while the second level represents (stations).
The practical application of the model shows that the factors that directly affect the increase in
deaths due to traffic accidents were the result of the driver. By (10.9617) on the number of
deaths, and that the variable of the vehicle (vehicle) affects by (8.9589) on the number of
deaths, and that the variable signals traffic light (7.9565) on the number of deaths. It is also clear from the results that (53%) of the differences in the estimates between stations
for the number of deaths are due to the random tendency of the driver variable, and (51%) of
the differences in the estimates due to the random tendency of the road variable, and (47%) of
the differences The estimates are due to the random inclination of the vehicle variable, and 44%
of the variations in the estimates are due to the random inclination of the variable light signals.
Key terms for research: traffic accident mortality, Poisson partial hierarchical regression
model, different inclination model, fullest possible place method.
It is also clear from the results that 53% of the differences in the estimates between stations
are due to the random tendency of the driver variable, 51% of the differences in the estimates
are due to the random tendency of the road variable, and 47% The estimates are due to the
random inclination of the vehicle's variable (the vehicle), and 44% of the differences in the
estimates are due to the random inclination of the light signal
variable.