Business Process Optimization Using Genetic Algorithm And Decision Trees

Abstract

Business process optimization is critical towards improving operational efficiency, reduction of costs and improved decision making in an organization. The current paper proposes a hybrid system comprising of Genetic Algorithms (GA) and Decision Trees (DT) to optimize features and parameters, model and extract decision rules, respectively. The difficulty in this case is to optimise these complicated business processes and still maintain the interpretability of the model which is required to make informed decisions. The GA determines the best parameters of business processes e.g. costs, time and resource allocation and the DT formulate a predictable model that can be understood and yields decision rules. The hybrid method exhibits better performance than the standalone models based on accuracy, precision, recall, and F1-Score. Experimental outcomes reveal that the accuracy, precision and recall of the proposed method are 89%, 87% and 88%, respectively, which is superior to the default Decision Tree and the GA optimized Decision Tree. Sensitivity analysis shows that the optimum size of the population is 50 and the optimum value of the mutation rate is 0.05. According to the confusion matrix, there are few false positives and false negatives of the model, meaning that the model is performing well. The results show the feasibility of hybrid GA+DT approaches for optimization of business processes in an interpretable manner. This is particularly applicable in circumstances where there is a need to have clear decisions made using rules. Subsequent studies may include real-time adaptation, increasing the volume of data sources, and connection with deep learning models to achieve more advanced performance.