The applications of real-time machine learning on human mobility data continue to grow, thanks to the increase in digital mobility data, such as phone records, GPS traces, and social media posts. This has led to exponential growth in areas such as next location prediction, crowd flow, trajectory generation, flow generation, disease spreading, urban projection, and well-being.
However, such predictions have many challenges such as combining multiple data sources from multiple devices in real-time, security and privacy measurements, the ability to apply machine learning on large-scale data without data loss or failovers, and most importantly, real-time stream processing to provide instantaneous reactions.
In this talk, you will learn how you can apply machine learning models to multiple human mobility datasets to provide real-time predictions, using Hazelcast, the Open-Source real-time data platform. The talk will address these challenges and provide solutions and best practices on how Hazelcast can optimize real-time machine learning predictions on human mobility datasets in the following areas: scalability, performance, failover, reliability, and data recovery.