Application Of Matrix Factor Model In Quantitative Analysis Of Regional Economy And Medical Image Recognition

High-dimensional panel data is widely present in many fields such as economics,finance,meteorology,and visual detection.Factor models are widely used in the era of big data for data dimensionality reduction.However,traditional factor models directly vectorize the data,ignoring the information about rows and columns in panel data.In order to more comprehensively and systematically study panel data and reveal the inherent rules of the data,matrix factor models have come into being.In January 2020,COVID-19 broke out in Hubei Province,China.Since then,the epidemic has wreaked havoc globally.Due to the virus’s rapid mutation and continuously increasing infectivity,China has achieved significant anti-epidemic results but also faces a long-term coexistence with the virus.In the post-epidemic era,how to coordinate the development of China’s regional economy is a key issue of concern for economic researchers.During the outbreak,how to use image recognition technology to assist doctors in efficiently diagnosing the disease and scientifically allocating medical resources is a matter of people’s immediate concern.This article examines the structural changes in China’s regional economic development pattern and medical image recognition based on the matrix factorization model.In the quantitative analysis of China’s regional economic development,this study first constructs a macroeconomic indicator system and verifies that regional economic data has a factor structure.Then,using the ACCE method,α-PCA method,and PE method of matrix factorization model,the factor loading matrix is estimated.The common factors are interpreted based on the factor loading,and the PE method is used to study the spatial and driving force structural changes in regional economies before and after the outbreak from a dynamic perspective.Finally,this article applies the matrix factorization model to rolling forecasts of the consumer price index(CPI),and the results show that the PE method achieves better predictive performance in most provinces than the ACCE and α-PCA methods.This article studies medical image recognition on human chest CT images that are labeled whether or not they have COVID-19.First,a matrix factorization model is used to estimate the common factors of the pixel matrix,and the estimated values of the common factors are used as features to train and test an image recognition model based on support vector machines,logistic regression,random forest.The study shows that the matrix factorization model can accurately extract the core structure of chest CT images and help the classifier accurately identify the images.After comparison,the PE method achieves higher F1 score and recall rate compared to the ACCE method and α-PCA method,and has certain advantages in the practical application of medical image recognition.