Introduction to Data Stream using River
Speaker: Tajudeen Akinosho
Track: Scientific Computing
Type: Remote Talk
Room: Talk Room 2 (Remote Talks)
Time: Oct 05 (Thu): 13:30
Data are continuously generated from diverse areas such data from new paradigms such as network devices, cloud computing, wireless and intelligent systems for communications, wireless sensor network (WSN), financial transactions, environmental monitoring, radio frequency identification (RFID), telecommunication, military surveillance, weather monitoring, Healthcare and medical diagnoses monitoring, electricity usage prediction, and real-time surveillance. In the streaming setting, data is unceasing, potentially unbounded, and continuously evolving with time. A tool that can be used as an extension on the Massive Online Analysis (MOA) framework is River, a Python library. This talk will introduce listeners how to use River to do machine learning on streaming data. River library supports different machine learning tasks, including regression, classification, and unsupervised learning. It also supports tasks such as computing online metrics and concept drift detection. River can be used to build a machine learning pipeline and evaluate the model’s performance on a streaming dataset.