Real Time Analytics

Course Code: MSCA 32005

Course Summary: Prerequisites:
MSCA 31006: Time Series and Forecasting
Students are expected to be comfortable enough with R to write software for processing and responding to streaming data.

One of the most actively developing areas of analytics is the real time analytics because of the growing number of data sources capable of collecting data round the clock in ever-larger amounts and with more complex structure; penetration of smart sensors everywhere where data collection used to be not possible, from micro to macro world and into hostile environments unsuitable for human observers; increasing demand for decisions made at latencies below human reaction time. Conducting real time analysis is different from the traditional data analysis in batch mode. Streaming data makes the very concept of sample nonexistent. Usual static sample characteristics, like p-value turn into dynamically changing processes. The old statistical concept of sufficient statistics may be getting a whole new meaning in the context of streaming data. The focus of the course is on stochastic methods suitable for real time analysis and their statistical implementations. Students will work with real data streaming live from the course server. We will learn about stochastic processes observed at random times and apply them to problems of monitoring, early event detection, prediction and control.