| Title |
Prediction Model for Preferred Color Temperature of Lighting User in Living Spaces Using Random Forest |
| Authors |
정재원(Jai-Won Chung) ; 방석오(Seok-oh Bang) |
| DOI |
https://doi.org/10.5370/KIEE.2024.73.10.1864 |
| Keywords |
LED Lighting; Color Temperature; Machine Learning; Multiclass Classification Model |
| Abstract |
In this paper, we propose the random forest model for prediction of the preferred color temperature of lighting user using usage environment data. Temperature, relative humidity, precipitation, wind speed, hour, AM/PM status, day of the week, weekday/weekend status, and color temperature setting were measured at one hour interval through theanthropism experiment, and the relation of each variable were investigated through the correlation coefficient. Random forest models were established using the 5-fold cross validation method, with approximately 80% of the total data (n = 231) as the training set (n = 184). The model performances were evaluated using the test set (n = 47), which accounts for approximately 20% of all the data. The random forest models could predict the preferred color temperature with a maximum accuracy of 74.5% using hour and temperature. |