Ground Water Quality Classification

This project aims to classify ground water quality using various machine learning models. The classification is based on a dataset containing ground water quality parameters for the years 2018, 2019, and 2020. Multiple models such as K-Nearest Neighbors, Logistic Regression, Support Vector Machine, Decision Tree, and Random Forest Classifier are used for classification tasks.

Dataset

The dataset consists of ground water quality data for three years:

• ground_water_quality_2018_post.csv
• ground_water_quality_2019_post.csv
• ground_water_quality_2020_post.csv

Data Preprocessing

The preprocessing steps include:

• Handling missing values by filling them with mean values.
• Renaming certain columns for consistency.
• Concatenating the datasets from three years into one.
• Transforming features using log transformation for better model performance.
• Encoding categorical variables into numerical values.

Exploratory Data Analysis (EDA)

EDA is performed to understand the distribution and relationships of various parameters in the dataset. Scatter plots and histograms are used to visualize the data.

Feature Engineering

The following features are selected for modeling:

• Latitude (lat_gis)
• Longitude (long_gis)
• Ground Water Level (gwl)
• pH
• Electrical Conductivity (E.C)
• Total Dissolved Solids (TDS)
• Carbonates (CO3)
• Bicarbonates (HCO3)
• Chlorides (Cl)
• Fluorides (F)
• Nitrates (NO3)
• Sulfates (SO4)
• Sodium (Na)
• Potassium (K)
• Calcium (Ca)
• Magnesium (Mg)
• Total Hardness (T.H)
• Sodium Adsorption Ratio (SAR)
• Residual Sodium Carbonate (RSC)

Classification Models

1. K-Nearest Neighbors (KNN)

Two KNN models are trained:

• Model 1: Classifies water quality based on multiple classes.
• Model 2: Classifies water quality into three classes: PS, US, MR.

2. Logistic Regression

Two Logistic Regression models are trained:

• Model 1: For multi-class classification.
• Model 2: For three-class classification (PS, US, MR).

3. Support Vector Machine (SVM)

Four different kernels are used for SVM:

• Linear
• Polynomial
• Radial Basis Function (RBF)
• Sigmoid

Grid Search with cross-validation is performed to find the best hyperparameters for each kernel.

4. Decision Tree Classifier

Two Decision Tree models are trained:

• Model 1: For multi-class classification.
• Model 2: For three-class classification (PS, US, MR).

5. Random Forest Classifier

Two Random Forest models are trained:

• Model 1: For multi-class classification with hyperparameter tuning.
• Model 2: For three-class classification (PS, US, MR) with hyperparameter tuning.

Usage

Prerequisites

• Python 3.x
• Libraries: pandas, numpy, matplotlib, seaborn, scikit-learn, google.colab

Steps to Run

1. Mount Google Drive:

   from google.colab import drive
   drive.mount('/content/drive')

2. Load the datasets:

   df_1 = pd.read_csv('/content/drive/MyDrive/project/ground_water_quality_2018_post.csv')
   df_2 = pd.read_csv('/content/drive/MyDrive/project/ground_water_quality_2019_post.csv')
   df_3 = pd.read_csv('/content/drive/MyDrive/project/ground_water_quality_2020_post.csv')

3. Preprocess the data as shown in the script.

4. Perform EDA using the provided visualizations.

5. Train the models:

   • KNN
   • Logistic Regression
   • SVM
   • Decision Tree
   • Random Forest

6. Evaluate the models using accuracy, confusion matrix, and classification report.

Model Evaluation

The models are evaluated using metrics such as accuracy, precision, recall, F1-score, and confusion matrix. The best parameters for each model are determined using Grid Search with cross-validation.

Results

The results of the models are visualized using heatmaps for confusion matrices and plots for accuracy scores. The detailed classification reports provide insights into the performance of each model.

Conclusion

This project demonstrates the application of various machine learning models for classifying ground water quality. The Random Forest model with 33 estimators showed the best performance for multi-class classification, while the SVM with the RBF kernel performed well for three-class classification.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Special thanks to the Google Colab platform for providing the environment to run this project.
• Data sourced from the ground water quality monitoring agencies.