ANALYSIS OF THE REGRESSION MODEL FOR ZERO-INFLATION DATA

Abstract

The community may contain a large percentage of zero values that cause the community distribution to move away from zero, and this group is referred to as not following a normal distribution, so one of the conditions of the linear regression models is permeated. This type of society can be seen in many applications such as insurance, meteorology, auditing, environment, and manufacturing. The zero-community number is often analyzed via a two-part admixture model: The first part is probabilistic from zero and the second part is regular with a specific probability distribution. Problems of confidence estimation of the zero-classifier population mean under normal models have been present in research. Regression models have also been developed for the zero population groups. However, many of these models are aimed at counting data, although regression models with responses of a continuous type can be seen in application quite often. Moreover, these regression models for homeless populations do not address situations in which the data available for analysis were obtained through complex probability sampling designs. Different statistical methods and models have been developed for the statistical analysis of such population. Based on the current research, most of the special studies focus on estimating the population mean and developing regression models. This dissertation will also focus on developing regression models. This dissertation will also focus on developing regression models. Most of the regression models developed for the null population found in research have given more attention to population data in which observations can take only non-negative integer values that arise from counting rather than ordering. They also use maximum possibility methods and pseudo greatest possibility methods to estimate expected responses in Value . Variable / future variables.