| Research Projects Supported by HKU's High Performance Computing Facilities |
| ˇ@ |
| Researcher: |
| Dr Philip L H Yu, Department of Statistics and Actuarial Science |
| Project Title: |
|
Spatial Models for Multiple Ranking Data and Their Applications |
| ˇ@ |
| Project Description: |
|
Ranking data are commonly seen in our daily life. In many business and social studies that involve comparing several items such as products, occupations and candidates, respondents are frequently asked to rank the items according to certain pre-defined criteria. Multiple ranking data are collected when (i) the same set of items are ranked in different occasions, one at a time, or (ii) two or more related sets of items are ranked separately. Very often, the main objectives in these studies are to identify the relationship among several sets of ranking and to explore the possible factors that influence individualsˇ¦ choice decisions. Over the past few years, spatial models including factor, vector and ideal-point models have been successfully developed to fit single ranking data. In this project, I shall develop new spatial models to fit multiple ranking data so as to provide a flexible framework to describe individual differences in various choice decisions. I will also focus on developing practically efficient estimation procedures for these spatial models. |
| ˇ@ |
| Project Duration: |
| 24 months |
| ˇ@ |
| Project Significance: |
| Understanding individual choice behavior is a topic of great interest to business analysts, market researchers, and social, educational and political policy makers. Marketing firms, product manufacturers, service providers, and many other organizations devote considerable resources to tackle this problem by conducting small to large-scale surveys. Frequently, multiple ranking data are obtained by asking respondents to (i) rank a set of items in different occasions, one at a time or (ii) rank two or more possibly related sets of items separately. This project aims at developing new statistical methodologies to analyze multiple ranking data. The analysis focuses on developing suitable spatial models for multiple ranking data to explain individual differences in various choice decisions. Current statistical investigations in business and social studies involving multiple ranking lack using the above sophisticated tools. I strongly believe that the new techniques will help a lot in many different areas. For instance, it can help market researchers to describe consumer behavior; help government officials to identify different views of employers on future manpower training after Chinaˇ¦s accession to World Trade Organization (WTO). |
| ˇ@ |
| Results Achieved: |
|
Yu, P.L.H. and Leung H.L. Bayesian analysis of wandering ideal-point models for displaying ranking data. In preparation. Yu, P.L.H. and Leung H.L. Wandering ideal-point models for multi-attribute ranking data. In preparation. One MPhil. Trained. |
| ˇ@ |
| Remarks on the Use of High Performance Computing Cluster: |
|
Carried extensive simulation studies, parallel programming and data analysis. |
| ˇ@ |
| Email Address: |
| plhyu@hku.hk |
| ˇ@ |