📈 LikeLogic-Engine

Predicting Facebook Post Engagement from Historical Interaction Signals

A regression-based machine learning system that predicts how many likes a Facebook post will receive, using engagement signals like unlikes, interactions, and video replays.

Live Demo

---

📌 Overview

LikeLogic-Engine predicts expected daily likes on a Facebook post using historical engagement patterns — enabling data-driven decisions about content performance before a post even goes live.

Built with Streamlit and a supervised Linear Regression model, it transforms raw engagement metrics (unlikes, interactions, video replays) into real-time predictive insight, trained on 34,000+ interaction records from Kaggle.

Features: engagement analytics dashboard · real-time like prediction · data insights on what drives performance

---

🏗️ Model

Algorithm	Linear Regression
Task	Supervised Regression
Input Features	Engagement metrics (unlikes/day, interactions, video replays)
Output	Predicted daily likes
Dataset	34,000+ interaction records (Kaggle)

📊 Performance

Fill in your eval numbers — R², MAE, and RMSE are the standard trio for regression.

R² Score	MAE	RMSE
—	—	—

---

🏛️ Architecture

User Input (Engagement Metrics) → Preprocessing (pandas/NumPy)
   → Feature Processing → Trained Model (Joblib - Linear Regression)
   → Prediction Output → Streamlit Visualization

---

🚀 Run Locally

git clone https://github.com/Pro-phet123/LikeLogic-Engine.git

cd LikeLogic-Engine

python -m venv venv

source venv/bin/activate      # Windows: venv\Scripts\activate

pip install -r requirements.txt

streamlit run social.py

---

🔮 Future Improvements

• Benchmark against Random Forest / XGBoost regressors
• Add feature importance visualization
• Expand inputs beyond likes/unlikes (e.g. comments, shares, post time)
• Add confidence intervals to predictions

---

👤 Author

Olalemi Olaoluwakintan Emmanuel — Data Scientist & AI Engineer

```