Scholarlink

<p>Food security in South Asia is a serious issue to be considered in the context of high population growth, climatic variation, and alteration in the structure of the economy. This paper gives a detailed, multi-method evaluation of the Food Production Index (FPI) in seven South Asian countries such as Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan, and Sri Lanka, as well as the aggregate of the region. Among these, non-parametric trend detection (Sens’ slope and Kendall’s tau), heatmap correlation, RF Machine Learning (ML) models with Leave One Out Cross Validation (LOOCV) compared to XGBoost and Lasso regression, SHAP-based feature importance analysis, and Error, Trend, and Seasonal (ETS)-based forecasting methods are employed for the period 2026–35. The empirical findings indicate that there are differences across the countries considered. Bangladesh, India, Nepal, Pakistan, and Sri Lanka have significant positive changes in FPI, whereas the Maldives has a significant negative change. The RF model performed better or equally well compared to other competing models, with R<sup>2</sup> ≥ 0.97 and R<sup>2</sup> ≥ 0.94 in the case of major agrarian economies (India, Bangladesh, Nepal, and Pakistan). RF performance is relatively worse in Bhutan and the Maldives, not due to a modelling weakness, but owing to structural volatility in their food production systems. Sri Lanka has shown moderate growth, which is vulnerable to changes in the economic and ecological environments. According to ETS forecasts, the growth rate will be slower for large agrarian economies by 2035, while for smaller and vulnerable states, the growth forecasts are associated with a high degree of uncertainty. The largest economies dominate regional food production, yet pronounced structural divergence in food security trajectories across South Asian countries, highlights critical gaps that aggregate indicators conceal, underscoring the urgent need for country-specific policy interventions. This study offers a novel methodological synthesis of statistical trend analysis, ML, and ETS forecasting, providing actionable insights for climate adaptation and the development of climate-resilient agricultural systems across South Asia.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Food security dynamics in South Asia: a multi-method analysis using Random Forest and Error Trend Seasonal models

  • 期刊论文

  • 发表日期：

    2026年7月8日

查看全文
  • Muhammad Atiq-ur-Rehman,
  • Sadia Mustafa,
  • Ayesha Hussain,
  • Irfan Hussain Khan,
  • Fatima Akhtar,
  • Sadaf Javed,
  • Azma Batool

摘要

Food security in South Asia is a serious issue to be considered in the context of high population growth, climatic variation, and alteration in the structure of the economy. This paper gives a detailed, multi-method evaluation of the Food Production Index (FPI) in seven South Asian countries such as Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan, and Sri Lanka, as well as the aggregate of the region. Among these, non-parametric trend detection (Sens’ slope and Kendall’s tau), heatmap correlation, RF Machine Learning (ML) models with Leave One Out Cross Validation (LOOCV) compared to XGBoost and Lasso regression, SHAP-based feature importance analysis, and Error, Trend, and Seasonal (ETS)-based forecasting methods are employed for the period 2026–35. The empirical findings indicate that there are differences across the countries considered. Bangladesh, India, Nepal, Pakistan, and Sri Lanka have significant positive changes in FPI, whereas the Maldives has a significant negative change. The RF model performed better or equally well compared to other competing models, with R2 ≥ 0.97 and R2 ≥ 0.94 in the case of major agrarian economies (India, Bangladesh, Nepal, and Pakistan). RF performance is relatively worse in Bhutan and the Maldives, not due to a modelling weakness, but owing to structural volatility in their food production systems. Sri Lanka has shown moderate growth, which is vulnerable to changes in the economic and ecological environments. According to ETS forecasts, the growth rate will be slower for large agrarian economies by 2035, while for smaller and vulnerable states, the growth forecasts are associated with a high degree of uncertainty. The largest economies dominate regional food production, yet pronounced structural divergence in food security trajectories across South Asian countries, highlights critical gaps that aggregate indicators conceal, underscoring the urgent need for country-specific policy interventions. This study offers a novel methodological synthesis of statistical trend analysis, ML, and ETS forecasting, providing actionable insights for climate adaptation and the development of climate-resilient agricultural systems across South Asia.