A Random Forest-based Model for Cargo Volume Prediction and Personnel Scheduling in Logistics Sorting Centers

The paper considers a problem of cargo quantity prediction and staff scheduling in logistics sorting centers. Depending on the different constraints existing in the problem, the random forest prediction model of cargo quantity is established, while the hybrid optimization algorithm is used to obtain the optimal solution for reasonable cargo quantity prediction and optimum attendance arrangement. Information processing at the stage of data processing was done with box plots and linear interpolation methods on the raw data. After that, a random forest prediction model was established for every sorting center's historical inventory data. The prediction of daily and hourly inventory data of 57 sorting centers in a logistics network for the next month was done based on this model. In the last phase, the objective function and constraint conditions had already been set up and were solved by a hybrid genetic algorithm according to the attendance plan model of regular and temporary employees. The results have shown that these two techniques can significantly forecast cargo volumes and, likewise, human resources scheduling, thereby offering reliable decision support to management within logistics sorting centers.