A data mining algorithm-based approach to accounting for enterprise operating costs

Abstract

Abstract This paper first proposes an enterprise operation costing method based on C4.5 optimization algorithm, which uses Taylor series to simplify the logarithmic operation and reduce the calculation time of the algorithm. The mean value of GINI index is used to eliminate the influence of inter-attribute correlation, dig deeper into the cost data, ensure the reliability and accuracy of the cost data, and thus improve the correct rate of cost accounting. Then the C4.5 algorithm and the C4.5 optimization algorithm are applied to the actual case of cost prediction and cost accounting of a city construction enterprise to compare the accuracy of the two algorithms. The dataset is used to iterate through the two algorithms to test and analyze their performance and global balance. The results show that the C4.5 optimization algorithm’s accounting value for each item is consistent with the actual value, and the accounting error is only 0.1%. The C4.5 algorithm has an accounting error rate of approximately 8%. The errors of both costing methods decreased to different degrees when the number of iterations increased from 10 to 100. The error of C4.5 algorithm decreases from 0.325 to 0.07. The error of C4.5 optimization algorithm decreases from 0.28 to 0.05. The error rate of the enterprise costing method of the C4.5 optimization algorithm is less, and the global balance is better.