Article

Risk Prediction and Influencing Factors Analysis of Poverty Monitoring Households —— based on Liangshan Yi Autonomous Prefecture, Sichuan Province

1 College of Mathematics and Computer Science, Guangdong Ocean University, Zhanjiang 524088, China
2 College of Economics, Guangdong Ocean University, Zhanjiang 524088, China

https://doi.org/10.58531/esmmsi/1/3/8

Received: 1 August 2024 / Accepted: 30 August 2024 / Published: 2 September 2024

This study utilizes Lasso Regression and MAHAKIL over-sampling to address multicollinearity and label imbalance issues in poverty-stricken household data from Liangshan Yi Autonomous Prefecture. Then a Bayesian-optimized LightGBM based on the TPE procedure was used for classification prediction and post-processing of the model using order-preserving regression. The results indicates that the accuracy of Xide increased by 6.62% and 1.35%, and the F1 score increased by 6.07% and 5.22%, respectively. Final, This study analyzes the impact of various factors on the risk of returning to poverty through decision plots, interaction maps, and waterfall charts focusing on per capita income, subsistence allowances, and working hours. This study not only demonstrates the important role of data mining in accurately identifying and preventing the recurrence of poverty, but also provides a scientific basis for the government and social organizations to help them formulate more effective poverty alleviation policies and development strategies.

Lasso Regression; Bayesian Optimization; Shapley Additive Explanations; Isotonic Regression; MAHAKIL

Hong Si., Li C., Huang J. Risk Prediction and Influencing Factors Analysis of Poverty Monitoring Households —— based on Liangshan Yi Autonomous Prefecture, Sichuan Province. Eng. Solut. Mech. Mar. Struct. Infrastruct., 2024, 1(3), doi: 10.58531/esmmsi/1/3/8

Back to TopTop