Plant Watering System
AI-Powered Plant Health & Watering Decisions
An AI-powered plant health monitor that predicts plant state from sensor and engineered features, then combines ML predictions with a rule-based engine to recommend watering actions. Built in Streamlit during an AI/ML fellowship.
Main Features
From sensor readings to a clear watering recommendation — with transparent ML and tunable rules.
Plant Health Prediction
Classifies each plant as Healthy, Needs Water, or Overwatered from sensor and engineered features.
Sensor & Feature Engineering
Soil moisture, temperature, and humidity signals cleaned and turned into model-ready features.
ML Model Benchmarking
Compares Logistic Regression, Random Forest, and XGBoost with accuracy, precision, recall, and F1.
Rule-Based Engine
Combines ML output with parameterized thresholds to recommend concrete watering actions.
Configurable Thresholds
Tune soil-moisture, temperature, humidity, and days-since-water thresholds via Settings sliders.
Validation & Leakage Checks
Stratified split + 5-fold CV and group-aware (Plant_ID) checks to catch data leakage.
Streamlit Dashboard
Interactive UI with prediction results, visualizations, and a settings page for live tuning.
Demo & Pipeline Visuals
Pipeline diagram, model-comparison figures, confusion matrices, and a short demo video.
Technology Stack
Build your ML product
From sensor data to deployable ML and Streamlit apps, we ship practical AI systems end to end.