Scholarlink

<p>Many switches and some thresholds occur frequently in most practical contexts of strategic decision making. Generally these two aspects have been studied separately. This study approaches the identification problem in a synthetic manner. Several combinations of objectives of decision making, with no clear combination in perspective, render it difficult to measure the dependent variable. A case in point is the concept of consumer loyalty to the products of a firm. Clearly, many independent variables have effects on an array of dependent variables.Three different methods of dealing with the unobservable dependent variable have been examined. In addition, almost all studies postulate that the switches and threshold levels satisfy a predefined probability distribution. The resulting estimates crucially depend on this assumption rather than reflect the nature of the data. By way of contrast, this study develops estimates of these changes without such an assumption. Further, suitable transformations of variables have been developed so that the non-linear regression can be converted into a simple linear regression. The case of 3MIndia was developed to illustrate the application of the proposed methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Efficient Algorithm for the Estimation of a Mixture of Switching and Threshold Regressions

  • 期刊论文

  • 发表日期：

    2025年3月19日

查看全文
  • T. V. S. Ramamohan Rao

摘要

Many switches and some thresholds occur frequently in most practical contexts of strategic decision making. Generally these two aspects have been studied separately. This study approaches the identification problem in a synthetic manner. Several combinations of objectives of decision making, with no clear combination in perspective, render it difficult to measure the dependent variable. A case in point is the concept of consumer loyalty to the products of a firm. Clearly, many independent variables have effects on an array of dependent variables.Three different methods of dealing with the unobservable dependent variable have been examined. In addition, almost all studies postulate that the switches and threshold levels satisfy a predefined probability distribution. The resulting estimates crucially depend on this assumption rather than reflect the nature of the data. By way of contrast, this study develops estimates of these changes without such an assumption. Further, suitable transformations of variables have been developed so that the non-linear regression can be converted into a simple linear regression. The case of 3MIndia was developed to illustrate the application of the proposed methods.