| Title |
A Process for Implementing Aging-friendly Smart Housing Model by Predicting Daily Activities |
| Authors |
주한나(Ju, Hannah) ; 이현수(Lee, Hyunsoo) |
| DOI |
http://dx.doi.org/10.14774/JKIID.2019.28.1.138 |
| Keywords |
Daily Activities ; Life-log ; Smart Environment ; Machine Learning ; Performance-Behavior Hybrid Model ; Dwelling Service |
| Abstract |
The purpose of this study is to propose a process that can provide the residential service by predicting the behaviors by time of the elderly using the life log data. In order to accomplish this purpose, first, the characteristics of daily life-log data and daily behavior patterns of elderly people were analyzed using the life-log data of previous studies which was collected through the 24 - hour self - filling questionnaire. Second, using the Weka program based on the collected life-log data, the behavior of the elderly by time of day was predicted and the accuracy of prediction was analyzed. Third, the process was simulated to provide the dwelling service in the elderly housing by extracting some of the predicted daily activities of elderly people and applying them to the Performance-Behavior Hybrid Model. A total of 1008 behaviors were analyzed. In order to propose a Performance-Behavior Hybrid Model, two behaviors (sleep, working out) with high prediction accuracy and two behaviors up washing, shower) behavior were extracted to provide light environmental dwelling service. The method was studied to provide dwelling services automatically using historical data of residents. In addition, it is meaningful that the possibility and the process of predicting the daily activities of the elderly through machine learning were explored. However, delicate techniques for detecting behavior and data collection that is large enough to improve prediction accuracy must be complemented. |