Research On Construction Project Tracking Audit Risk Based On Bayesian Network

With the continuous expansion of the scale of government-invested construction projects,tracking audit has become one of the main audit modes of government-invested construction projects in order to effectively monitor the quality of projects and the use of funds.However,there are many problems in the implementation of tracking audit,such as long implementation time,various types of professional contacts and participation units in the implementation process.Therefore,there are many risk factors,which may lead to tracking audit risks and fail to achieve the desired results.In order to strengthen process control,reduce tracking audit risk and improve audit quality,this paper introduces interpretative structural model and Bayesian network to provide new ideas for research and management of tracking audit risk of construction projects.Based on literature search and expert interview,this study identifies 22 risk factors of construction project tracking audit from four aspects:audit object,audit subject,audit environment and other risks,such as incomplete data of project proposal and feasibility report,insufficient examination of bidding qualification,imperfect laws and regulations of tracking audit,excessive number of tracking audit projects,insufficient professional ability of auditors and lack of experience.On this basis,the expert group of 10 people’s interpretation structure model is established,and the construction project tracking audit risk adjacency matrix is established.The Python code is used to solve the reachability matrix,and the construction project tracking audit risk interpretation structure model is constructed,which lays the foundation for the construction of Bayesian network model.Through the questionnaire survey to obtain data,the use of risk matrix will be standardized questionnaire data into GeNIe 2.1 Academic Bayesian network structure learning,the formation of a complete construction project tracking audit risk Bayesian network model.Then the Bayesian network parameter learning is carried out by using the data.Through the diagnostic reasoning,sensitivity analysis and the most approximate factor chain analysis of the Bayesian network model,it is concluded that:①frequent replacement of auditors-poor management continuity,insufficient professional ability of auditors-lack of experience,auditors ’ participation in project management and audit information transmission risk are the key factors for construction project tracking audit risk.②Eight factors,including imperfect laws and regulations of follow-up audit,insufficient professional competence of auditorslack of experience,lack of audit standards and systems,frequent replacement of auditorspoor management continuity,auditors participation in project management,excessive number of follow-up audit projects,low professional ethics quality of auditors,and weak risk awareness of auditors,are the factors that have high risk sensitivity of follow-up audit of construction projects.③The incomplete data of project proposal and feasibility report,the inconsiderate examination of bidding qualification,the imperfect laws and regulations of follow-up audit,the excessive number of follow-up audit projects and the lack of professional ability and experience of auditors are the five causes of the risk of follcw-up audit of construction projects.A total of nine causal chains of follow-up audit of construction projects are derived from these five causes.Then,according to the actual data of the NJCJ bridge project and the interview with the participants,the risk level of each factor is comprehensively judged,and it is input into the Bayesian network model of the construction project tracking audit risk.It is concluded that the NJCJ bridge project tracking audit risk is small,which is consistent with the facts,and the validity of the Bayesian network model of the construction project tracking audit risk is verified.Finally,according to the above analysis,targeted suggestions are put forward to reduce the risk of construction project tracking audit and ensure the quality of tracking audit.In this study,the interpretative structural model and Bayesian network are used to truly reflect the interaction between construction project tracking audit risk,identify the main factors of construction project tracking audit risk and make recommendations,broaden the existing research methods of construction project tracking audit risk evaluation and increase the direction of risk management,which has certain theoretical and practical significance.