Designing data pipelines for analytics and machine learning in industrial settings

Machine learning has made it possible for technologists to do amazing things with data. Its arrival coincides with the evolution of networked manufacturing systems driven by IoT. In this presentation we’ll examine the rise of IoT and ML from a practitioners perspective to better understand how applications of AI can be built in industrial settings. We'll walk through a case study that combines multiple IoT and ML technologies to monitor and optimize an industrial heating and cooling HVAC system. Through this instructive example you'll see how the following components can be put into action: 1. A StreamSets data pipeline that sources from MQTT and persists to OpenTSDB 2. A TensorFlow model that predicts anomalies in streaming sensor data 3. A Spark application that derives new event streams for real-time alerts 4. A Grafana dashboard that displays factory sensors and alerts in an interactive view By walking through this solution step-by-step, you'll learn how to build the fundamental capabilities needed in order to handle endless streams of IoT data and derive ML insights from that data: 1. How to transport IoT data through scalable publish/subscribe event streams 2. How to process data streams with transformations and filters 3. How to persist data streams with the timeliness required for interactive dashboards 4. How to collect labeled datasets for training machine learning models At the end of this presentation you will have learned how a variety of tools can be used together to build ML enhanced applications and data products for instrumented manufacturing systems. Speakers IAN DOWNARD Sr. Developer Evangelist MapR WILLIAM OCHANDARENA Senior Director of Product Management MapR