title	🌿 HaveCrops Analytics
emoji	🚜
colorFrom	green
colorTo	blue
sdk	streamlit
pinned	false

🌾 HaveCrops Analytics: AI Advisory System

An End-Semester Capstone Project: From Predictive Analytics to Intelligent Intervention

HaveCrops Analytics is an advanced, production-grade agricultural engine. It transitions the agricultural ecosystem from traditional machine learning yield prediction (Milestone 1) into a fully automated, Agentic Farming Advisor operating on LangGraph, RAG (Retrieval-Augmented Generation), and LLM pipelines (Milestone 2).

---

🏗️ System Architecture & Milestones

This repository completes the full lifecycle of the agronomic advisor architecture:

Milestone 1: Predictive Analytics

• Classical Machine Learning: Predicts exact crop yields based on soil, rainfall, and fertilizer geometries using Scikit-Learn pipelines (Linear Regression, Decision Tree CART).
• Continuous Visual Dashboard: Plots interactive multi-variable analytics connecting predictive outcomes natively with historical distributions via Plotly.
• Geochemical Risk Metrics: Flags real-time thresholds tracking yield probabilities mapped through exact data logic.

Milestone 2: LLM Agentic Advisory (Current)

• LangGraph Node Routing: Reasons dynamically regarding farm environments.
• RAG (Retrieval-Augmented Generation): Synthesizes vector embeddings of agronomic best-practices directly into the local context.
• Formatted Generative Reports: Securely exports programmatic step-by-step PDF solutions outlining explicitly modeled action plans dynamically bound to the user's soil health predictions!

---

🎨 Premium Dark-Mode User Interface

The entire frontend has been reconstructed into a SaaS-Level Dark Dashboard, natively configuring:

• Overview Matrix: 4-Node card grids explicitly guiding user interaction workflows.
• Visual Analytics Engine: Interactive Plotly distributions dynamically updating on continuous threshold sliders.
• System Architecture Diagrams: Native HTML visualizations of the complex AI inference pipeline embedded securely inside the application.
• Dynamic Risk Scaling: Gauges inversely bound to mathematical dataset metrics.

---

⚙️ Technical Stack

Layer	Technology
Prediction Engine	Scikit-Learn
Statistical Analysis	Pandas, NumPy, SciPy
Interactive Visualizations	Plotly (px, go)
Intelligence Layer	LangGraph, FAISS + sentence-transformers (RAG), Groq (Llama 3.3)
Interface & DOM	Streamlit (Custom Theme Architecture)

---

🚀 Running the App Locally

To clone the deployment natively and test the code workflows:

1. Create a virtual environment and assemble dependencies:

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

2. Establish the Groq Network Key: To securely generate the AI agentic protocols, create a .env file right inside the root folder, and paste your LLM key:

   GROQ_API_KEY="your-groq-key-here"

   Note: This file is fully protected by .gitignore and securely prevents your private key from leaking online.

3. Launch the Engine:

   streamlit run app.py

(For live cloud deployments like Vercel or Streamlit Cloud, inject your API keys directly into the web "Environment Variables" / "Secrets" dashboard instead of utilizing .env!)

---

🎓 Academic Deliverables & Evaluation

This platform aligns fully as an End-Semester submission for Advanced Machine Learning in Agriculture.

• Track: BTech CSE | AI & ML
• Scale: End-to-End ML Pipeline + LLM Agent Workflow Implementation.

👥 Team Members

Name	PRN	Role
Vedant Satbhai	2401010500	Team Leader
Pratik Agade	2401010346	Report Maker & UI Checker
Raghvendra Singh	2401010367	Research Analytics & UI/UX

Evaluation parameters heavily target System Contextual Routing, Vector Retrieval Stability, Interface Precision, and Deployment Operations UX constraints mapped extensively throughout this repository.

Status: Submission Ready ✅