ALCOHOL-RELATED CATEGORICAL VARIABLES--PHASE II

约束分类回归(CCR)方法最初是 在一个成功的第一阶段项目中开发。的长期目标是 这项第二阶段的提案是作为用户制定CCR方法-- 友好的软件包，自然地将先验知识整合为 逻辑因果关系直接转化为范畴关系分析 因变量数据。这样的统计工具将大大增加 通过使更有效的方法来了解与酒精相关的疾病 利用现有和未来的机构、县、州和国家 水平数据库。 二期工程的具体目标是：1)利用CCR 酒精研究数据的分析方法，2)开发了一个基于Unix的 界面友好的CCR软件包，以及3)进一步细化一期CCR 方法通过额外的算法设计和开发。因此， 预期的第二阶段成果将包括：1)更好地了解 一个特定的与酒精有关的问题通过应用 CCR建模方法；2)商品化的原型CCR建模 软件产品。因为将先验知识融入到 在流行病学、临床和社会研究领域，CCR模型具有 作为商业数据成功开发的巨大潜力 分析产品。

The Constrained Categorical Regression (CCR) methodology was originally developed in a successful Phase I project. The long-term objective of this Phase II proposal is to develop-the CCR methodology as a user- friendly software package that naturally integrates prior knowledge as logical causal relationships directly into the analysis of categorical dependent-variable data. Such a statistical tool will greatly increase current knowledge of alcohol-related conditions by making more effective use of existing and future institutional, county, state, and national level data bases. The specific aims of the Phase II project are: 1) utilize the CCR approach to analyze data from alcohol research, 2) develop a UNIX-based user-friendly CCR software package, and 3) further refine the Phase I CCR methodology through additional algorithm design and development. Thus, the expected Phase II results will include: 1) a better under- standing of a particular alcohol-related problem obtained through application of the CCR modeling approach, and 2) a marketable prototype CCR modeling software product. Because the incorporation of prior knowledge into the areas of epidemiologic, clinical, and social research, CCR modeling has tremendous potential for successful development as a commercial data analysis product.