This Proof of Concept (POC) aimed to demonstrate the feasibility of using machine learning and social media sentiment analysis to forecast taxi demand and location hotspots across Dubai. The project sought to integrate real-time social media data with historical taxi usage patterns to improve operational efficiency and resource allocation.
Business Objectives
- Improve taxi availability in high-demand areas through proactive demand forecasting.
- Reduce passenger wait times and optimize fleet distribution.
- Leverage public sentiment and event triggers on social platforms (e.g., X (former Twitter), Instagram) to anticipate surges in demand.
An intelligent assembly of widely spread fragmented and non-semantic data can be aggregated by latest technology advancements to alleviate some of the widely spread customer problems.
Solution Approach
MWB combined historical taxi ride data, socia media feeds and weather and event data available by using technologies such as Natural Language Processing (NLP) for sentiment analysis (using libraries like spaCy, TextBlob, BERT), Facebook Prophet for time-series forecasting, Geospatial Analysis to map demand clusters, and ETL pipelines for integrating APIs for social media ingestion.
Project Outcome and Benefits
This POC successfully demonstrated how social sentiment and real-time signals can enhance traditional forecasting models for public transport demand. The next steps involved expanding data coverage, integrating more real-time feeds, and deploying a scalable production system for city-wide taxi management. Interesting observations were made through demand forecasting dashboard and demonstrated against the actual data points.