An empirical study on finding experience sampling parameters to explain sleep quality based on dimension reduction

The ground truth which is given a set of sensor data from life-long data whose definition are ambiguous, it is important to develop the way self-reporting without user intervention. The imperatively necessary items of self-reporting with the minimum number can be one solution. Therefore, we suggest the method that reduce the number of items of self-reporting with investigate representative items of others. Then we collect the data in real life and evaluate our proposed method. Before data collection, the items of self-reporting have been designed exquisitely for investigating critical activities and context for sleep quality. We then designed an experiment that collects both passive sensing data for sleep quality and self-reported data for daily activities to investigate activities and context closely relate to sleep quality. We reduced the number of items of self-reports in a manner of exploratory data analysis on this data and evaluate the results through logistic regression. The results indicate that a small number of items which is shown to be important can be considered for consisting items of self-reporting.