Search

Zhimu Huanxi: A Custom Big Data & Machine Learning Platform to Quantify Air Pollution Impacts on Regional Livestock Economic Output

Author(s)

Lu Li, Jingyang Cui, Jiaqi Zhang*

Affiliation(s)

School of Information Science and Technology, Baotou Teachers' College, Baotou, Inner Mongolia, China *Corresponding Author

Abstract

Farmers and local environmental regulators lack practical tools to quantify how ambient air pollutants drag down livestock industry revenue. To fill this practical gap, our team built a customized analysis platform named Zhimu Huanxi, using multi-source field records collected across Inner Mongolia between 2020 and 2024. The raw datasets cover local meteorological observations, routine air quality monitoring records and annual livestock production statistics. After clearing invalid records, aligning time labels and normalizing all numerical indicators, we adopted Lasso regression to screen out six dominant impact factors: heavy pollution coverage intensity, atmospheric SO₂ concentration, inhalable PM10 particle levels, regional annual average temperature, livestock breeding scale and sustained mild pollution periods. We then constructed a Linear Support Vector Regression (LSVR) model to predict yearly livestock output value losses triggered by air pollution. We reserved all 2024 records as independent test samples to test generalization capacity; the model reached an R² of 0.78, with an average absolute forecasting error of 152 million RMB and root mean square error equal to 198 million RMB. Residual values scattered randomly around zero without systematic deviation. The platform backend runs on lightweight Flask, while the front-end embeds ECharts interactive charts. Operators can view pollutant weight rankings, real-time output loss forecasts and multi-year historical data with one-click query. This platform delivers quantifiable evidence for breeders to design emission reduction plans and helps environmental authorities lock high-priority pollution control zones in pastoral areas.

Keywords

Machine Learning; Multi-Source Big Data; Air Pollution Economic Assessment; Livestock Output; Lasso Feature Screening; Linear Support Vector Regression

References

[1] Xie Q J, Zheng P, Bao J, et al. Temperature and Humidity Prediction Model in Closed Pig House Based on Deep Learning. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(10): 353-361. [2] Liu C H, Yang L, Deng H, et al. Prediction of Ammonia Concentration in Pig House Based on ARIMA and BP Neural Network. China Environmental Science, 2019, 39(6): 2320-2327. [3] Yang L, Liu C H, Guo Y C, et al. Prediction of Ammonia Concentration in Pig House Based on EMD-LSTM. Transactions of the Chinese Society for Agricultural Machinery, 2019, 50(S1): 353-360. [4] Park J, Jo G, Jung M, Oh Y. Comparative Analysis of Neural Network Models for Predicting Ammonia Concentrations in a Mechanically Ventilated Sow Gestation Facility in Korea. Atmosphere, 2023, 14(8): 1248. [5] Genedy R A, Chung M, Shortridge J E, et al. A physics-informed long short-term memory (LSTM) model for estimating ammonia emissions from dairy manure during storage. Science of the Total Environment, 2024, 912: 168885.