Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.



Published on

  • Be the first to comment

  • Be the first to like this


  1. 1. - machine learning for machine data TL;DL - Build predictive models on data streams automatically - Generate actionable predictions in real time - Integrates into your application infrastructure: JSON in, JSON out - Looking for early adopters in finance, operations, monitoring
  2. 2. (C) 2013 Analytics today “Analytics” eg Data warehouse OLAP cubes Dashboards ? Action Event streams
  3. 3. (C) 2013 Analytics today “Analytics” eg Data warehouse OLAP cubes Dashboards ? Action Event streams Problems: - Queries defined in advance - you have to know what you’re looking for - Difficult to integrate into application infrastructure - humans watching dashboards Solution: - Computers are good at spotting patterns - give the humans a hand - Mine for patterns automatically
  4. 4. (C) 2013 Event streams Real-time actions Featurestream stream mining engine Streaming machine learning, incremental model building Enriched events Predictions Anomalies ...
  5. 5. (C) 2013 Real-time actions { id:10, time:1022391, type:SIGV, value:92.1 } { id:10, time:1022391, type: {actual:SIGV, expected:SIGV, confidence:0.93}, value: {actual:92.1, expected:108.2, confidence:0.82} } Enriched events Predictions Anomalies ... Event streams Featurestream stream mining engine
  6. 6. (C) 2013 Event streams Data warehouse OLAP cubes Dashboards Enriched events Predictions Anomalies ... Real-time actions Packaged into a complete stream analytics solution Featurestream stream mining engine
  7. 7. (C) 2013 Example: infrastructure monitoring • Monitor metrics of web service • Featurestream learns typical behaviours • A disk has mechanical failure • Featurestream alerts you that • Latency of response has increased • The disk has higher response times • You can quickly identify the fault
  8. 8. (C) 2013 Use cases • Finance: • Predicting prices/movements from prior prices and public news feeds (e.g. Reuters, BBC news, social media) • Finding anomalous events in trading data • Operations/monitoring: extracting signals from syslog/splunk/click data • Smart metering: real-time signals from smart meter data • Marketing: real-time features from social media/news streams
  9. 9. (C) 2013 Next steps • Get an access key at and try it (hosted service) • Talk to us about adding streaming machine learning into your application infrastructure