|
|
Short-term Agricultural Operating Risk Prediction with Big Data of Ecommerce:A Case Study of Family-run Pear Farmers |
CHENG Xinwei, YUE Zhonggang |
School of Economics, Nanjing University of Posts and Telecommunications |
|
|
Abstract: Agricultural e-commerce has a positive effect on operation improvement and welfare enhancement. Identifying and predicting its risks helps to actively manage and intervene embedding patterns, and also provides theoretical basis and early warning solutions for regional e-commerce optimization. Electronic ledger data of production, transaction and finance of 3755 pear farmers in Jiangsu Province were collected to compare the predictive capability of linear model, 1-dimensional non-linear machine learning models and 2-dimensional deep learning model on the risk outcomes and decisions of agricultural operations. The results show that ledger data have adequate information content to predict short-term agricultural risk. Convolution Neural Networks on feature grey scale has the best prediction ability. And the prediction information of transaction data is higher than production and financial data with mutually validated structure. Recommendations are made from the perspectives of e-commercial data specification, ledger software upgrading, consulting service construction and digital platform management.
|
Received: 14 April 2022
Published: 15 September 2022
|
|
|
|
|
|
|
|