摘要
数字经济规模是衡量数字经济发展水平的量化指标。在参考和梳理数字经济规模影响因素基础上,收集整理江西省2011~2020年的指标数据,通过对样本缺失数据采用线性回归进行月份多重插补扩充,采用相关性分析及多重共线性分析实现冗余指标删除,通过主成分分析实现数据降维,同时构建基于数据处理的PCA-BPNN机器学习预测模型及对比模型,对江西省数字经济规模预测进行实证分析。实验结果表明:PCA-BPNN模型预测结果MAE、MAPE分别为240.0181、0.0233,预测精度相较于构建的对比模型均提高了30%以上,最高达73.8%,论证了基于数据处理的机器学习预测模型的有效性以及准确性。该方法为区域科学制定数字经济发展战略具有重要的理论与现实价值。
The scale of the digital economy is a quantitative indicator to measure the level of development of the digital economy. On the basis of referring to and sorting out the factors affecting the scale of the digital economy, collect and sort out the indicator data of Jiangxi Province from 2011 to 2020, use linear regression for monthly multiple imputation expansion of sample missing data, use correla-tion analysis and multi-collinearity analysis to delete redundant indicators, use principal compo-nent analysis to achieve data dimensionality reduction, and build a PCA-BPNN machine learning prediction model and comparison model based on data processing. Conduct empirical analysis on the prediction of the scale of digital economy in Jiangxi Province. The experimental results show that the MAE and MAPE predicted by the PCA-BPNN model are 240.0181 and 0.0233, respectively. Compared with the constructed comparative model, the prediction accuracy has been improved by more than 30%, with a maximum of 73.8%. This demonstrates the effectiveness and accuracy of the machine learning prediction model based on data processing. This method has important theoreti-cal and practical value for regional science in formulating digital economy development strategies.
出处
《应用数学进展》
2023年第6期2915-2923,共9页
Advances in Applied Mathematics
