Presented By O'Reilly and Cloudera
Make Data Work
Dec 4–5, 2017: Training
Dec 5–7, 2017: Tutorials & Conference
Singapore

Querying time series patterns with SAX

Supreet Oberoi (Oracle)
2:35pm3:15pm Thursday, December 7, 2017
Average rating: ****.
(4.00, 1 rating)

Who is this presentation for?

  • Data scientists, chief IoT officers, and VPs of analytics

Prerequisite knowledge

  • A basic understanding of the MapReduce paradigm

What you'll learn

  • Learn how Oracle adapted and extended symbolic aggregate approximation (SAX) to the MapReduce paradigm and applied it to solve problems in the oil and gas, manufacturing, healthcare, and automobile industries

Description

Time series data is any dataset that is plotted over a range of time (market data (stock prices over time), EKG (volts over a period of time), etc.). Often, in IoT use cases, what is of interest is finding a pattern in the sequence of measurements. For example, there may be a pattern of pressure readings before an oil pipe ruptures. The operator may want to track that pattern in real time to prevent future ruptures or search the historical database to do root-cause analysis. However, queries on time series data do not traditionally scale. To unleash the potential of big data for IoT-centric use cases, we must adapt ad hoc and predictive analytics to patterns of data

Supreet Oberoi explains how Oracle adapted and extended symbolic aggregate approximation (SAX) to the MapReduce paradigm and applied it to solve problems in the oil and gas, manufacturing, healthcare, and automobile industries. Specifically, Supreet explores the seminal query patterns required for time series analytics and how to develop digital signatures that allow for efficient comparison for equality between patterns. Along the way, Supreet shares techniques for doing fuzzy matches, since in time series data, no two patterns are exactly alike. You’ll learn how Oracle tested these algorithms at scale and see how they provide the foundation for real-time analysis to detect anomalies and even causation and correlation across different time series dimensions.

Photo of Supreet Oberoi

Supreet Oberoi

Oracle

Supreet Oberoi is vice president of engineering, IoT, and big data at Oracle. A technology executive with over 20 years of experience building products and solutions for real-time, distributed, and big data analytical applications, Supreet has held technical and leadership roles at Oracle, Concurrent, American Express, Real-Time Innovations, Agile, Microsoft, and many other privately held Silicon Valley companies. He is also the lead mentor for Stanford student startup accelerator startX. Supreet holds a BS in computer sciences with highest honors from the University of Texas at Austin and an MS in computer sciences from Stanford University. He is widely published and often presents at conferences. In his free time, Supreet is reconnecting with an old passion—painting.