APPLICATION OF INTELLECTUAL ANALYSIS TOOLS FOR SOLVING WAREHOUSE OPERATION OPTIMIZATION PROBLEMS

Abstract

This work is devoted to the currently relevant issues - optimization of warehousing. Cargo receiving and dispatch processes held in a medium-sized warehouse are being considered in this work in detail. Based on the results of preliminary analysis, several indicators were selected, necessary for the usage of intellectual analysis tools in order to predict the required number of employees at the warehouse entrances. In accordance with the purpose set, models for predicting the required number of employees at the entrances were created to ensure the working process in a way, which corresponds to the optimal value of the indicator “workload of the entrance”. Various methods of machine learning, such as decision tree, k-nearest neighbors regression, random forest, and feedforward neural network are considered in the context of the problem, mentioned above. Each of the models was trained with different values of the model's hyperparameters, which were selected in both manual heuristic-based mode and using specialized software tools for grid search (GridSearchCV) from the scikit-learn library, designed to find the optimal values of the hyperparameters. Using the automated search for hyperparameters when training models yields to a smaller mean-square error in comparison with manual selection of hyperparameters. According to the analysis results of the model prediction quality, it was found that the predicted number of employees closely corresponds to the real situation in comparison with the planned values being used by the company. Based on the obtained results, several recommendations were made to assess the growth of the economic efficiency of the enterprise.