"Marketing spend is wasted on the edge of a cliff. We predict the fall."
TrendFall AI is a next-generation forensic analysis platform designed for CMOs and digital strategists. It identifies the exact "Cringe Point" of digital trends before they collapse, allowing brands to exit high-risk campaigns and reallocate capital with surgical precision.
Frontend: https://trendfall-ai-yz54.vercel.app/ Backend API: https://trendfall-backend.onrender.com
Leverages SHAP (SHapley Additive exPlanations) to provide transparent, multi-dimensional justifications for every risk score. No black boxes—just pure forensic data.
Real-time time-series forensic tracking of trend momentum. Identify whether a trend is in a Healthy Growth, Plateau, or Terminal Decay phase.
A specialized radar-visualization that tracks audience burnout across four key dimensions:
- Content Saturation Degree
- Sentiment Degradation
- Engagement Momentum Volatility
- Natural Maturity Index
Adjust "What-If" parameters like Slope, Sentiment, and Fatigue to see how your ROI would shift in different market conditions.
Compare two competing trends side-by-side to determine which asset deserves the majority of your marketing budget.
| Layer | Technologies |
|---|---|
| Frontend | React 19, Vite, Tailwind CSS, Framer Motion, Recharts |
| Visuals | Three.js, React Three Fiber, GLSL Shaders |
| Backend | Python 3.10+, FastAPI, Uvicorn |
| Machine Learning | Scikit-Learn, SHAP (Explainable AI), Pandas, NumPy |
| Neural Brain | Google Gemini 1.5 Flash (GenAI) |
| Data Sources | YouTube Data API v3, SerpAPI (Google Trends & News) |
├── frontend/ # React + Vite + Tailwind (Deployed on Vercel)
│ ├── src/components/ # Glassmorphic UI Components
│ └── src/App.tsx # Main Dashboard Orchestrator
├── backend/ # FastAPI + ML Engine (Deployed on Render)
│ ├── main.py # API Endpoint Layer
│ ├── trend_engine.py # Main Analysis Pipeline
│ ├── ml_model.py # Risk Prediction Logic
│ └── genai_explainer.py# Decision Justification Engine
└── .env # Secrets Configuration (Local only)
- Python 3.10+
- Node.js 18+
git clone https://github.com/Jalmesh53/Datathon-2026.git
cd Datathon-2026cd backend
pip install -r requirements.txt
python main.pycd frontend
npm install
npm run devIn the current attention economy, the difference between a high-ROI campaign and a brand-damaging failure is often just 48 hours. TrendFall AI provides the forensic "early warning system" needed to stay ahead of the curve.
Created for the 2026 Datathon.