School of Humanities and Social Science Division of Social Science 198 The Political Economy of Conflict and Elections Supervisor: HENDRY David James / SOSC Student: LUI Wing Lam / QSA Course: UROP 1000, Summer As the number of the non-English speakers in United States is growing, the ballots in some of the states would provide the bilingual version of ballot for the voters of language minority to better understand the policy making. Since the language minority is less likely to fully understand the ballot measure in English compared with their first language, therefore, it is crucial to investigate the effect of bilingual ballot on the language minority’s political support. This project will be gathered the documentary of the ballot archive, ballot measure and perform analysis to estimate the language minority support on the ballot measures correlate with the bilingual ballot in different states. The Political Economy of Conflict and Elections Supervisor: HENDRY David James / SOSC Student: WONG Ming Pan / GCS-SOSC Course: UROP 1100, Fall UROP 2100, Spring This report serves as a follow-up report to the previous progress report with regards to the progress of the dataset analyses about the characteristics and feasibility of the datasets, especially in terms of the data availability and quality of the IPUMS dataset and some concerns about the IPUMS datasets with alignment to the ultimate conflictual event datasets, to clarify and set up the precondition that the framework or approach we would need for repetitive work before and in the poststratification methods in estimating the geographic distribution of subjective grievances to conflictual events. Modelling Sleep-related Data Using Hidden Markov Models Supervisor: HSIAO Janet Hui-wen / SOSC Student: GAO Shuyang / MATH-STAT LUO Hongfei / MATH-STAT Course: UROP 1100, Summer UROP 1000, Summer In this study, we applied the EMHMM toolbox to analyze physical activity data using Hidden Markov Models. Our objective was to uncover temporal patterns and state transitions within the data. We employed both one-dimensional and two-dimensional approaches, revealing that a single group with multiple states provided the best model fit. Clustering analysis showed that while partitioning into two groups was feasible, a cohesive group of all subjects was more effective. The findings suggest significant insights into activity behaviors, and future work could integrate sleep data to explore the relationship between physical activity and sleep patterns.
RkJQdWJsaXNoZXIy NDk5Njg=