| Title |
Occupancy Detection Model in Living Spaces using Random Forest and Indoor Environment Measurement Data |
| DOI |
https://doi.org/10.5573/ieie.2021.58.9.58 |
| Keywords |
Occupancy detection; Indoor environment measurement data; Logistic?regression; Random forest |
| Abstract |
Few studies have looked at the occupancy detection methods in living spaces using machine learning methods and indoor environment measurement data. In this paper, we propose the random forest model for occupancy detection using temperature, humidity, CO2, and illuminance values. Temperature, relative humidity, CO2, illuminance, and occupancy status were measured at thirty seconds interval through the experiment, and the characteristics and relation of each variable were investigated through the T-test and the point-biserial correlation coefficient. About 80% of the total measured data (n = 43,904) was used as the training set data (n = 35,123). The logistic regression model using the backward elimination method and the random forest model using the 5-fold cross validation method were constructed using the training set data, respectively. About 20% of the total data was used as the test set data (n = 8,781) for evaluation of the models. The accuracy of the logistic regression model using four values was 78.2%. The accuracy of random forest model using four values were 98.7%, and the accuracy of the model using only illuminance value was 81.2%. |