NateNjogu/Water-Quality-and-Monitoring-System

A distributed machine learning model that predicts and monitors the quality of water from sources across Kenya # 🇰🇪 Predictive Modeling of Water Quality in Kenyan Streams ## Project Overview This project is an offline-capable, edge-computing predictive modeling system designed to assess water safety in rural Kenyan streams. It utilizes **PySpark** for big data processing and **TensorFlow Lite** for decentralized, offline inference on mobile devices and laptops. ## Authors * Nathan Karoki * Emmanuel Ngeti ## Key Features * **Big Data Processing:** Uses PySpark to ingest, clean, and normalize large-scale regional environmental data. * **Deep Machine Learning:** A Sequential Neural Network trained to map complex chemical interactions (pH, Turbidity, Conductivity) to binary safety classifications (MAE: 0.0846). * **Edge Computing:** The model is compressed into a `.tflite` format, allowing field agents to make instantaneous water safety predictions without an internet connection. * **Streamlit Dashboard:** An interactive UI for filtering regional data and conducting manual offline predictions. ## Repository Contents * `Water Quality Monitor (1).ipynb`: The PySpark data pipeline and TensorFlow model training code. * `purity_edge_model.tflite`: The compressed offline Edge AI model. * `water_quality_app.py`: The Streamlit dashboard application. * `Project_Summary_Guide_No_IoT.pdf`: Complete system documentation. * `2_Design_Mockups_Milestone1.pdf`: UI design wireframes. ## How to Run the Dashboard Locally 1. Ensure you have Python installed. 2. Install the required libraries: `pip install streamlit pandas numpy` 3. Run the application: `streamlit run water_quality_app.py`