Scholarlink

The problem of safe drinking water is the biggest challenge faced by Leh, which is the northernmost part of India. Leh City is facing a problem of non-potable water and water scarcity. To address this problem, there is a need for effective water harvesting techniques and water quality management. This study examines water quality parameters: pH, temperature, potassium, calcium, chloride, magnesium, total hardness, and alkalinity for total dissolved solids prediction. The proposed framework utilizes interpolation techniques to increase the dataset population and investigate relevant features using mutual information importance and recursive feature elimination. A machine-learning model called Random Forest is used to predict the total dissolved solids. Random forest with RFE achieved an \(R^2\) value of 0.991 with a lower MAE of 0.0128 and an MSE of 0.0003, outperforming random forest with mutual information, which achieved an \(R^2\) value of 0.990 with a lower MAE of 0.014 and an MSE of 0.0004. This finding indicates that the Random Forest model with recursive feature elimination is more suitable for the Leh Ladakh data set.

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

Feature Selection and Classification Using an Ensemble Learning Model on Leh Ladakh Water Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nitesh Kumar Sahu,
  • Mridu Sahu,
  • Roopa Golchha,
  • D. C. Jhariya,
  • Mayank Shrivastav,
  • Chandan Kumar Singh

摘要

The problem of safe drinking water is the biggest challenge faced by Leh, which is the northernmost part of India. Leh City is facing a problem of non-potable water and water scarcity. To address this problem, there is a need for effective water harvesting techniques and water quality management. This study examines water quality parameters: pH, temperature, potassium, calcium, chloride, magnesium, total hardness, and alkalinity for total dissolved solids prediction. The proposed framework utilizes interpolation techniques to increase the dataset population and investigate relevant features using mutual information importance and recursive feature elimination. A machine-learning model called Random Forest is used to predict the total dissolved solids. Random forest with RFE achieved an \(R^2\) value of 0.991 with a lower MAE of 0.0128 and an MSE of 0.0003, outperforming random forest with mutual information, which achieved an \(R^2\) value of 0.990 with a lower MAE of 0.014 and an MSE of 0.0004. This finding indicates that the Random Forest model with recursive feature elimination is more suitable for the Leh Ladakh data set.