Scholarlink

<p>This research is employed for the applicability of artificial intelligence and machine learning for the prediction by application of Random Forest Regressor (RFR) model to predict fluoride concentrations in water samples, utilizing key water quality parameters. The performance of the model is assessed using several metrics: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Nash-Sutcliffe Efficiency (NSE), and Coefficient of Determination (R²). The RFR model performs with MSE of 0.0068, RMSE of 0.0825 and NSE and R² of 0.863, which implies it performs with good predictive accuracy. It is also observed that in the learning curve that the model suffered from overfitting, as evidenced by the distance between training and validation errors, which indicates good performance in training data. Feature importance analysis demonstrates that chloride, sodium, magnesium, and potassium variables are the most important predictors of fluoride, while factors such as electrical conductivity and pH are shown as moderately important factors. These results validate the effectiveness of machine learning model for predicting fluoride concentrations in groundwater to address water contamination issues.</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

Applicability of Random Forest Regressor for Prediction of Fluoride in Groundwater

  • 期刊论文

  • 发表日期：

    2025年4月30日

查看全文
  • Abhishek Kumar Mishra,
  • Nityanand Singh Maurya

摘要

This research is employed for the applicability of artificial intelligence and machine learning for the prediction by application of Random Forest Regressor (RFR) model to predict fluoride concentrations in water samples, utilizing key water quality parameters. The performance of the model is assessed using several metrics: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Nash-Sutcliffe Efficiency (NSE), and Coefficient of Determination (R²). The RFR model performs with MSE of 0.0068, RMSE of 0.0825 and NSE and R² of 0.863, which implies it performs with good predictive accuracy. It is also observed that in the learning curve that the model suffered from overfitting, as evidenced by the distance between training and validation errors, which indicates good performance in training data. Feature importance analysis demonstrates that chloride, sodium, magnesium, and potassium variables are the most important predictors of fluoride, while factors such as electrical conductivity and pH are shown as moderately important factors. These results validate the effectiveness of machine learning model for predicting fluoride concentrations in groundwater to address water contamination issues.