Research on Cargo Volume Forecasting and Manpower Demand Based on LSTM and Linear Programming Models

The research on cargo volume forecasting and manpower demand involves utilizing appropriate models and algorithms to predict future information based on historical cargo volume and personnel allocation data, thereby enhancing resource optimization and driving development. After preprocessing the data provided by a certain enterprise, this study utilizes LSTM (Long Short-Term Memory) patterns with reference to the daily cargo volume of each sorting center over the past four months and the hourly cargo volume over the past 30 days. A stationary time series model is constructed, and LSTM neural network parameters are set. Through rolling forecasts conducted via Python programs, an effective prediction of the hourly cargo volume for each of the 57 sorting centers over the next 30 days is achieved. Subsequently, a linear programming model is established to determine the objective function for maximizing personnel arrangement efficiency. Constraints such as single-shift attendance and employee headcount limits are selected. The Branch and Bound algorithm is employed to initialize upper and lower bounds and determine the feasible space, effectively yielding personnel deployment plans for different time periods at each sorting center over the next 30 days. The forecasting methods and data obtained in this paper play a significant role in resource allocation and service quality improvement in related industries.