DATA SCIENCE - Traffic Congestion Recommendations System Based On Climate Condition - 100% Complete

[20127304-AQ]

Week 10 Pull Request for review of Sprint 3-5

Changes made:

• Updated full clean dataset after preprocess from 2 initial dataset.
• Updated README.md with Dataset Description, Data Preprocessing, Feature Engineering, Exploratory Data Analysis, Machine Learning Models, Model Fine-Tuning, LLM Explanation System, System Optimization, Conclusion, Future Improvements
• Updated notebook for Individual Use Case - Traffic Congestion Recommendations System

Use case implementation details:

• Optimize model performance with TimeSeriesSplit and finetuning RandomizedSearchCV.
• Evaluate models using Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics before and after fine-tuning.
• Analyze feature importance to identify key environmental and temporal congestion factors.
• Use LLM prompt engineering to generate human-readable traffic explanations.
• Integrate RAG with traffic knowledge rules using TF-IDF and cosine similarity retrieval to enhance system response.
• Generate context-aware travel recommendations and suggested travel times.
• Combine ML predictions with AI-generated insights for smart transportation support.
• Support urban mobility planning and improve transportation efficiency and sustainability.