Scholarlink

<p>In this paper, the saline groundwater issues is identified in the basaltic hard rock type, which is a crucial for socio-economic and health risk assessment in the Akola and Buldhana districts of Maharashtra in India. Therefore, to measuring the salinization of groundwater resources is vital to safeguarding the safety of groundwater resources and ecological management. The primary objectives of this research are to develop machine learning (ML) models to predict the sodium percentage (Na%) in irrigation water, thereby understanding the irrigation water quality index and assessing health risks due to sodium. Therefore, both scenarios were selected based on different combinations of input variables for predicting sodium percentage (Na%) by using subset regression models. In this study, two scenario models were compared in terms of accuracy using various statistical performance metrics to select the best model in each scenario. The results of both scenarios, the first scenario shows the GBR model is high performance compared with other models, which is achieved accuracy measured in training and testing phases such as (Training: R<sup>2</sup>= 0.88, RMSE = 2.98, <i>R</i> = 0.94, MSE = 8.92, MAE = 2.37, and MARE = 15.08, Bias = 0.00 and Testing: R<sup>2</sup>= 0.70, RMSE = 6.55, <i>R</i> = 0.85, MSE = 42.94, MAE = 5.47, and MARE = 36.89, Bias = 0.64). In the second scenario the results of the study demonstrated that the GBR model fits better than the other models, results also shows the GBR model results better performance as compared with other models, the accuracy estimations of two phases such as (Training: R<sup>2</sup>= 0.87, RMSE = 3.15, <i>R</i> = 0.94, MSE = 9.95, MAE = 2.50, MARE = 15.58, Bias = 0.00 and Testing: R<sup>2</sup>= 0.71, RMSE = 6.46, <i>R</i> = 0.86, MSE = 41.79, MAE = 5.27, MARE = 35.27, Bias = 0.74). Based on the Bias, the GBR models shows a moderate and balanced bias in the first and second scenarios, also showing the lowest RMSE with higher R<sup>2</sup> values. RF models shows higher R² and lower RMSE values in both scenarios; however, both scenarios are shows highest bias values, hence this model is not superior as compared with Bagging models. Therefore, considering the overall performance across all evaluation metrics, the GBR model was identified as the most effective and reliable model in both scenario. The hazard quotient (HQ) is considered for computed non-carcinogenic health risk assessment concerns linked with groundwater drinking. In this study, we have included the non-carcinogenic HQ above 1 for sodium ingestion in the studied wells has acquired from whole locations. The sodium health risk assessment and prediction of sodium percentage was combinely studied for both drinking as well as irrigational purposes in the study area. The study’s results could provide support for regional, national, and local experts in enhancing groundwater development and planning. This area is important for the view of irrigation purses and agricultural production, such as cotton and , and in turn national economy. </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

Prediction of sodium percentage index for irrigation purpose using decision tree, and boosting models with health risk assessment

  • 期刊论文

  • 发表日期：

    2026年5月28日

查看全文
  • Kanak N. Moharir,
  • Neyara Radwan,
  • Fahad Alshehri,
  • Jitendra Rajput,
  • Abhay M. Varade,
  • Malay Pramanik,
  • Chaitanya Baliram Pande

摘要

In this paper, the saline groundwater issues is identified in the basaltic hard rock type, which is a crucial for socio-economic and health risk assessment in the Akola and Buldhana districts of Maharashtra in India. Therefore, to measuring the salinization of groundwater resources is vital to safeguarding the safety of groundwater resources and ecological management. The primary objectives of this research are to develop machine learning (ML) models to predict the sodium percentage (Na%) in irrigation water, thereby understanding the irrigation water quality index and assessing health risks due to sodium. Therefore, both scenarios were selected based on different combinations of input variables for predicting sodium percentage (Na%) by using subset regression models. In this study, two scenario models were compared in terms of accuracy using various statistical performance metrics to select the best model in each scenario. The results of both scenarios, the first scenario shows the GBR model is high performance compared with other models, which is achieved accuracy measured in training and testing phases such as (Training: R2= 0.88, RMSE = 2.98, R = 0.94, MSE = 8.92, MAE = 2.37, and MARE = 15.08, Bias = 0.00 and Testing: R2= 0.70, RMSE = 6.55, R = 0.85, MSE = 42.94, MAE = 5.47, and MARE = 36.89, Bias = 0.64). In the second scenario the results of the study demonstrated that the GBR model fits better than the other models, results also shows the GBR model results better performance as compared with other models, the accuracy estimations of two phases such as (Training: R2= 0.87, RMSE = 3.15, R = 0.94, MSE = 9.95, MAE = 2.50, MARE = 15.58, Bias = 0.00 and Testing: R2= 0.71, RMSE = 6.46, R = 0.86, MSE = 41.79, MAE = 5.27, MARE = 35.27, Bias = 0.74). Based on the Bias, the GBR models shows a moderate and balanced bias in the first and second scenarios, also showing the lowest RMSE with higher R2 values. RF models shows higher R² and lower RMSE values in both scenarios; however, both scenarios are shows highest bias values, hence this model is not superior as compared with Bagging models. Therefore, considering the overall performance across all evaluation metrics, the GBR model was identified as the most effective and reliable model in both scenario. The hazard quotient (HQ) is considered for computed non-carcinogenic health risk assessment concerns linked with groundwater drinking. In this study, we have included the non-carcinogenic HQ above 1 for sodium ingestion in the studied wells has acquired from whole locations. The sodium health risk assessment and prediction of sodium percentage was combinely studied for both drinking as well as irrigational purposes in the study area. The study’s results could provide support for regional, national, and local experts in enhancing groundwater development and planning. This area is important for the view of irrigation purses and agricultural production, such as cotton and , and in turn national economy.