발간년도 : [2022]
| 논문정보 |
|
| 논문명(한글) |
[Vol.17, No.4] Comparison of Forecasting Models for Fire Damage Based on Machine Learning - Focusing on Property and Human Damage |
|
| 논문투고자 |
Gyoung-Bae Kim, Weonil Jeong |
|
| 논문내용 |
In spite of various efforts on laws and systems for fire-fighting and technological approaches for fire detecting and preventing to minimize fire damage, both property damage and human damage have not been reduced. Recently, machine learning technology is being used in various intelligent application services, and its use is gradually increasing in the field of fire prevention and response. Accordingly, several kinds of research were conducted to prepare the basis for fire prevention and response by predicting the risk or possibility of a fire using machine learning technologies such as big data-based RNN(Recurrent neural networks), RF(Random forest), and boosting. Our study conducted a comparative evaluation to analyze and predict property damage and casualties caused by fire using ARIMA(Autoregressive integrated moving average), random forest, and LSTM(Long short term memory) models in order to present the direction of application of machine learning techniques in research for predicting fire damage. In this paper, the prediction accuracy of the target machine learning method for property damage and personal damage was evaluated using RMSE(Root mean square error) as the selection criterion. From the experimental results, it was found that the LSTM(Long short term memory) model was optimal for both the prediction of property damage and human damage. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
