SocialProphet

Hybrid Time-Series Forecasting & Generative Content Agent

A system that bridges the gap between social media engagement prediction and content generation by creating an integrated "Predict → Generate" pipeline.

Overview

SocialProphet uses time-series forecasting models to analyze historical engagement data, then leverages these insights to generate context-aware content recommendations through a generative AI agent.

Core Innovation

Unlike existing tools that either predict engagement OR generate content, SocialProphet creates an actionable bridge between analytics and content creation.

Features

• Time-Series Forecasting: Ensemble approach using Prophet + SARIMA
• Insight Extraction: Automated pattern recognition and trend analysis
• Content Generation: LLM-powered content recommendations via Hugging Face
• FIIT Framework: Quality validation (Fluency, Interactivity, Information, Tone)

Project Structure

SocialProphet/
├── data/
│   ├── raw/              # Original datasets
│   ├── processed/        # Cleaned data
│   └── predictions/      # Model outputs
├── notebooks/
│   ├── 01_EDA.ipynb      # Exploratory analysis
│   ├── 02_Forecasting.ipynb
│   ├── 03_Insights.ipynb
│   └── 04_Generation.ipynb
├── src/
│   ├── data_processing/  # Data collection & preprocessing
│   ├── forecasting/      # Time-series models
│   ├── insights/         # Pattern extraction
│   ├── generation/       # LLM integration
│   ├── evaluation/       # Metrics & visualization
│   └── utils/            # Configuration & helpers
├── tests/                # Unit tests
├── dashboard/            # Streamlit app
├── requirements.txt
└── setup.py

Installation

1. Clone the repository:

git clone https://github.com/avkbsurya119/SocialProphet.git
cd SocialProphet

2. Create virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up environment variables:

cp .env.example .env
# Edit .env with your API keys

Quick Start

from src.forecasting.prophet_model import ProphetForecaster
from src.generation.content_gen import ContentGenerator

# Initialize the pipeline
forecaster = ProphetForecaster()
generator = ContentGenerator()

# Train and predict
forecaster.train(historical_data)
predictions = forecaster.predict(days_ahead=7)

# Generate content recommendations
recommendations = generator.generate(predictions)

Technology Stack

• Forecasting: Facebook Prophet, SARIMA (statsmodels)
• Generation: Hugging Face Inference API (Llama 3.1)
• Data Processing: Pandas, NumPy, Scikit-learn
• Visualization: Matplotlib, Seaborn, Plotly
• Dashboard: Streamlit

Evaluation Metrics

• Forecasting: RMSE, MAE, MAPE, R² Score
• Content Quality: FIIT Framework compliance
• System: End-to-end latency < 5 seconds

Contributors

• navadeep555
• avkbsurya119

License

MIT License

Acknowledgments

• Facebook Prophet team
• Hugging Face community
• Academic references cited in the project report