基於駕駛風險等級建立決策樹模型以探討國道客運偏差駕駛行為之管理

A Decision Tree Based on Driving Risk Levels to Manage the Aberrant Behavior of Highway Bus Drivers

國道上的交通事故易導致嚴重的人員傷亡與財物損失。尤其國道客運駕駛員須面對長時間且高風險的行駛，每次偏差駕駛行為的發生都可能帶來難以承擔的後果，安全管理更為重要。許多研究都曾探討事故成因，卻鮮少評估風險等級。另外，事故未發生並不代表零風險，應藉其他指標衡量風險，以達預防之效。本研究考量各偏差駕駛行為之於風險的重要性相異，引入客觀賦權法進行風險評級，並以決策樹來直觀且清楚地展示駕駛行為對風險等級的影響，提供客運公司有效之風險管理工具。管理人員僅需依循該公司已建立的決策樹模型，即能簡易地評估駕駛員的風險等級，據以執行適切的改善措施。本研究建立之模型顯示，未保持安全距離是區別高風險駕駛員最關鍵之行為。

Traffic crash occurred on freeway often leads to serious casualties and property damage. A highway bus driver must pay full attention in a high-risk working environment during their career. Each time aberrant driving behavior occurred may result in severe consequences. Hence risk management has always been valued by bus companies. Although many studies have investigated the causes of crashes, few have assessed drivers' risk level to establish a rigorous accident prevention system. In addition, zero accident does not equal safety. We need to measure risk by other indicators to prevent accidents. This study grades risk by an objective weighting method considering different importance to risk of various types of aberrant driving behaviors. Moreover, decision tree analysis is applied to display the impact of driving behaviors on risk level, providing bus companies with intuitive risk management tool. The risk level of any driver can be graded effectively through calibrated tree models. These models established in this study show that violation of safety distance is the key behavior to distinguish high-risk drivers.