TY - GEN
T1 - Boosting Methods for Predicting Firemen Interventions
AU - Cerna, Selene
AU - Guyeux, Christophe
AU - Arcolezi, Héber H.
AU - Royer, Guillaume
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/4
Y1 - 2020/4
N2 - Forecasting future incidents to the next hours is of great importance for fire brigades, which allows improving their response time to interventions, one direct cause to guarantee proper attendance to victims. Moreover, for many departments around the world, it exists problems such as the high increment of interventions through the years, which requires more personnel and machinery resources. However, due to budget limitations fire brigades have to face this increment with the same resources. Therefore, the objective of this paper is to implement and compare three boosting methods in the specific task of predicting the number of firemen interventions in the next hour. A dataset with specific temporal information of interventions from 2006-2018 was provided by the department fire and rescue SDIS25 in Doubs-France. Great efforts were concentrated on arranging and collecting more data (e.g., meteorological data, road traffic conditions). As it is presented in this work, they were prepro-cessed and learned in a supervised way. As shown in results, such methods are mature enough to provide a good solution with an acceptable margin of error for real-life implementations.
AB - Forecasting future incidents to the next hours is of great importance for fire brigades, which allows improving their response time to interventions, one direct cause to guarantee proper attendance to victims. Moreover, for many departments around the world, it exists problems such as the high increment of interventions through the years, which requires more personnel and machinery resources. However, due to budget limitations fire brigades have to face this increment with the same resources. Therefore, the objective of this paper is to implement and compare three boosting methods in the specific task of predicting the number of firemen interventions in the next hour. A dataset with specific temporal information of interventions from 2006-2018 was provided by the department fire and rescue SDIS25 in Doubs-France. Great efforts were concentrated on arranging and collecting more data (e.g., meteorological data, road traffic conditions). As it is presented in this work, they were prepro-cessed and learned in a supervised way. As shown in results, such methods are mature enough to provide a good solution with an acceptable margin of error for real-life implementations.
KW - AdaBoost
KW - Firemen
KW - Gradient Boosting
KW - Prediction
KW - XGBoost
UR - https://www.scopus.com/pages/publications/85084989528
U2 - 10.1109/ICICS49469.2020.239488
DO - 10.1109/ICICS49469.2020.239488
M3 - Contribution to conference proceedings
AN - SCOPUS:85084989528
T3 - 2020 11th International Conference on Information and Communication Systems, ICICS 2020
SP - 1
EP - 6
BT - 2020 11th International Conference on Information and Communication Systems, ICICS 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 11th International Conference on Information and Communication Systems, ICICS 2020
Y2 - 7 April 2020 through 9 April 2020
ER -