With the rapid development of big data technology, the number of economic data grows faster, and it is more challenging for people to grasp and calculate economic data. In this paper, an economic statistical model is proposed to optimize data-cleaning technology. The economic data application framework is constructed using Multi-Agents, and the mining of economic data is achieved using Microsoft time series and clustering algorithm. The GMDH algorithm, which is the core of self-organized data mining, is proposed and improved to complete economic data analysis and prediction by using the fitting error or prediction variance criterion as the identification criterion. The wild value identification method based on the regression model is utilized to eliminate potential wild values and carry out data cleaning. Statistical analysis of the economic data of the printing industry in Shanghai, China, reveals that the total assets, total industrial output value, and total profit of the printing industry in Shanghai in 2023 declined by 2.63%, 4.77%, and 5.68%, respectively. External investment and enterprise R & D investment up to 109,440,100 U.S. dollars, 140,301,000 yuan. The overall number of employees declined, and the profit margin on output value decreased by 5.68%. It is predicted that the number of enterprises, total assets, total industrial output value, and total profit of Shanghai’s printing industry will rebound in 2024, and the external direct investment and R&D investment will be appropriately reduced.