https://github.com/sunnyrao07/water-quality-analysis

A machine learning project that predicts water potability based on chemical and physical attributes, using models like Logistic Regression, Random Forest, and XGBoost.
https://github.com/sunnyrao07/water-quality-analysis

data-cleaning label-encoding logistic-regression matplotlib model-evaluation numpy pandas pyhton random-forest sckiit-learn seaborn smote standard-scaler xgboost

Last synced: 20 days ago
JSON representation

A machine learning project that predicts water potability based on chemical and physical attributes, using models like Logistic Regression, Random Forest, and XGBoost.

• Host: GitHub
• URL: https://github.com/sunnyrao07/water-quality-analysis
• Owner: SunnyRao07
• Created: 2025-04-13T13:55:53.000Z (22 days ago)
• Default Branch: main
• Last Pushed: 2025-04-13T14:05:34.000Z (22 days ago)
• Last Synced: 2025-04-13T14:41:04.595Z (22 days ago)
• Topics: data-cleaning, label-encoding, logistic-regression, matplotlib, model-evaluation, numpy, pandas, pyhton, random-forest, sckiit-learn, seaborn, smote, standard-scaler, xgboost
• Language: Jupyter Notebook
• Homepage:
• Size: 0 Bytes
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# 💧 Water Potability Analysis

## 📌 Project Overview  

Water quality is a pressing global issue, and access to safe drinking water is essential for health and sustainability. This project aims to **predict the potability of water** samples using **machine learning models** based on chemical and physical attributes.  

Three classification models were implemented and evaluated:  

- **Logistic Regression**  

- **Random Forest Classifier**  

- **XGBoost Classifier** (🏆 Best Performing)

---

## 📂 Project Resources  

🔹 **Dataset (CSV File)**: [Download water_potability.csv](https://github.com/SunnyRao07/Water-Quality-Analysis/blob/main/water_potability.csv)  

🔹 **Project Code (.ipynb)**: [View Jupyter Notebook](https://github.com/SunnyRao07/Water-Quality-Analysis/blob/main/Water_Quality_Analysis.ipynb)  

🔹 **Project Report (DOCX File)**: [Download Report](https://github.com/SunnyRao07/Water-Quality-Analysis/blob/main/Water%20Quality%20Analysis.docx)

---

## 🧾 Dataset Overview  

- **Total Records**: 3,276  

- **Target Variable**: `Potability` (1 = Safe to drink, 0 = Unsafe)  

- **Features**:

  - `ph`

  - `Hardness`

  - `Solids`

  - `Chloramines`

  - `Sulfate`

  - `Conductivity`

  - `Organic_carbon`

  - `Trihalomethanes`

  - `Turbidity`

📝 **Class Imbalance Notice**: Only ~39.8% of the samples are potable, requiring resampling strategies like **SMOTE**.

---

## 🛠 Data Preprocessing  

✔️ **Handling Missing Values**  

- Imputed missing values in `ph`, `Sulfate`, and `Trihalomethanes` using **median imputation**

✔️ **Feature Scaling**  

- Standardized numerical columns using **StandardScaler**

✔️ **Class Imbalance Handling**  

- Applied **SMOTE (Synthetic Minority Over-sampling Technique)** to balance potable and non-potable classes

✔️ **Outlier Detection**  

- Identified and handled outliers using visualizations and statistical techniques

---

## 📊 Exploratory Data Analysis  

- **Visualization Tools**: Histograms, box plots, scatter plots, heatmaps  

- **Key Insights**:  

  - `Turbidity`, `pH`, and `Conductivity` have the most influence on potability  

  - `Solids` and `Conductivity` are highly correlated  

  - Certain features have skewed distributions and outliers

---

## 🤖 Models Used  

### 1️⃣ Logistic Regression  

✔️ Simple and interpretable baseline  

❌ Less accurate on complex patterns  

### 2️⃣ Random Forest Classifier  

✅ Robust and interpretable  

✅ Handles non-linear relationships well  

❌ Slower on large datasets  

### 3️⃣ XGBoost Classifier (🏆 Best Model)  

✅ High accuracy and performance  

✅ Efficient with imbalanced datasets  

✅ Strong generalization and feature importance ranking

---

## 📈 Model Performance

| Model               | Accuracy | Precision | Recall | F1 Score |

|--------------------|----------|-----------|--------|----------|

| Logistic Regression| ~63.5%   | ~59.3%    | ~49.7% | ~53.9%   |

| Random Forest       | ~74.2%   | ~70.1%    | ~67.5% | ~68.8%   |

| **XGBoost**         | **78.6%** | **75.3%**  | **73.8%** | **74.5%**  |

📌 **XGBoost** performed best across all evaluation metrics.

---

## 🔍 Feature Importance (XGBoost)  

Top contributors to potability prediction:  

- `Turbidity` 🥇  

- `pH`  

- `Conductivity`  

- `Chloramines`  

- `Trihalomethanes`

---

## 🚀 Future Scope  

🔹 **Integrate IoT Sensors** for real-time data collection  

🔹 **Add microbial and temperature data** for improved predictions  

🔹 **Explore deep learning models** for more complex patterns  

🔹 **Deploy as a web app** for public and government use

---

## 🏗 Project Contributions  

This project includes the following key components:  

✔️ Exploratory Data Analysis  

✔️ Data Cleaning & Preprocessing  

✔️ Model Training, Tuning, and Evaluation  

✔️ Report and Code Documentation  

---