Longitudinal emotion analysis with GLOBEM dataset

Date

2024-9

Author

Koçoğlu, Elif Yeşim

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

96
views

62
downloads

Cite This

The Positive and Negative Affect Schedule (PANAS) is a widely used psychometric scale to measure positive and negative affect. This study uses GLOBEM mobile sensing data from the Pyhsionet database to analyze the Positive and Negative Affect Schedules (PANAS) scores. For this analysis, classical statistical methods such as marginal and linear mixed models and machine learning approaches, including MERF, MERT, and REEMtree, along with their stochastic variants, are used. Our results indicate that models incorporating random effect terms outperform the marginal model. In particular, as shown by the R2 results, MERF shows the best prediction performance on the training dataset, while (S)MERF performs well on the test dataset. These results suggest that individual traits substantially impact future emotional states more than factors such as sleep, physical activity, and phone use.

Subject Keywords

Longitudinal data analysis, Positive and negative affect schedule, Machine learning models, Marginal models, Linear mixed models

URI

https://hdl.handle.net/11511/111323

Collections

Graduate School of Natural and Applied Sciences, Thesis