Cetas Analytics as a Service for Predictive Analytics


Published on

Predictive Analytics

Published in: Technology
  • Be the first to comment

Cetas Analytics as a Service for Predictive Analytics

  1. 1. Predictive Analytics Leads to SuccessfulRecommendations and Direct Revenue MaximizationChristos Tryfonas, Ph.D.CTO, Cloud & Big Data Analytics INSTANT INTELLIGENCEWeb:  www.cetas.net  Twi)er:  @CetasAnaly/cs  Blog:  www.cetas.net/blog  YouTube:  www.youtube.com/CetasAnaly/cs   © 2009 VMware Inc. All rights reserved
  2. 2. INSTANT INTELLIGENCE Transform IT + Transform Business + Transform Yourself Leverage Cetas to Transform Real-Time and Predictive Analytics2
  3. 3. Infrastructure, Apps and now Data… INSTANT INTELLIGENCE Build Run Private Public Analyze ManageSimplify Infrastructure Simplify App Platform Simplify Data With Cloud Through PaaS with (IaaS) Analytics as-a-Service 3
  4. 4. Trend: New Data Growing at 60% Y/Y INSTANT INTELLIGENCEExabytes of information stored 20 Zetta by 2015 1 Yotta by 2030 Yes, you are part of the yotta audio   generation… digital  tv   digital  photos   camera  phones,  rfid   medical  imaging,  sensors   satellite  images,  logs,  scanners,  twi)er   cad/cam,  appliances,  machine  data,  digital  movies   Source: The Information Explosion, 20094
  5. 5. Data Growth in the Enterprise INSTANT INTELLIGENCE5
  6. 6. Trend: Value from Data Exceeds Hardware Cost INSTANT INTELLIGENCE§  Value from the intelligence of data analytics now outstrips the cost of hardware •  Hadoop enables the use of 10x lower cost hardware •  Hardware cost halving every 18 months Value Big Iron: $40k/CPU Commodity Cluster: $1k/CPU Cost 6
  7. 7. The Big Data Problem INSTANT INTELLIGENCE Velocity Volume Variety Value $ 10’s of Billions From Terabytes to Multi-Structured Businessof Daily Records Petabytes Insights7
  8. 8. Trend: Big Data Analytics – Driven by Real-World Benefit INSTANT INTELLIGENCE8
  9. 9. Big Data Use Cases INSTANT INTELLIGENCE9
  10. 10. Big Data Analytics ROI INSTANT INTELLIGENCESource: Nucleus Researchhttp://www.enterpriseappstoday.com/business-intelligence/the-more-pervasive-business-analytics-roi-big-data.html 10
  11. 11. Big Data Technology Stack INSTANT INTELLIGENCE11
  12. 12. A Holistic View of a Big Data System INSTANT INTELLIGENCE Real-Time Streams Real-Time Processing AnalyticsIngestion & Big Query, ETL Real Time Search, Batch Database (HBase, Index Processing Cassandra) (Google, Spunk, Cetas) Unstructured Data (HDFS)12
  13. 13. The Unified Analytics Cloud Platform INSTANT INTELLIGENCE Mahout R Cetas/VMware Analytics Tools Splunk SAS or IBM SPSS Map Reduce Developer Spring PaaS Python Frameworks Cloud Foundry Cassandra HBase HDFS Database/DataStores Greenplum Voldemort Hadoop Data Platform Data PaaS VMware /Cetas vSphere Cloud Infrastructure Private Public13
  14. 14. A Unified Analytics Cloud Significantly Simplifies Infrastructure INSTANT INTELLIGENCE §  Simplify •  Single Hardware Infrastructure •  Faster/Easier provisioningSQL or NoSQL Cluster Big SQL NoSQL Hadoop Analytics Engine Analytics L Cluster Unified Analytics Infrastructure Private Public Hadoop Cluster §  Optimize •  Shared Resources = higher utilization Decision Support Cluster •  Elastic resources = faster on-demand access 14
  15. 15. Predictive Analytics Objectives INSTANT INTELLIGENCE§  Optimization of one or more target functions •  Example: Increase Customer Engagement§  Vertical Specific •  Increase Customer Satisfaction •  Maximize Customer/User Engagement •  Minimize Churn Rate •  Maximize Revenue15
  16. 16. Sample Applications of Predictive Analytics INSTANT INTELLIGENCE§  eCommerce •  Customer interaction •  Revenue Maximization Statistical / •  Inventory management Regression§  Mobile Models •  Capacity Planning§  Ad/Publishing •  Ad serving Machine •  Customer Retention Management Learning (NN, Bayesian,§  Gaming etc.) •  Ad serving •  Virtual Good Placement •  In-App Purchasing Social Graph§  IT Algorithms •  Capacity Planning •  Alerting and Diagnosis16
  17. 17. Taking Predictive Analytics to logical conclusion INSTANT INTELLIGENCE§  Currently predictive analytics is in Silos •  Data Scientists w/ SAS, SPSS, Custom models •  Localized/Sampled Datasets •  Building Models using Open Source Modeling Tools: Hadoop, Mahout, R§  Actionable Analytics •  Recommendation Engines (Examples: Amazon, Netflix) •  Product Placement Engines •  Targeted Ad Placement Engines§  Cloudification of Predictive Analytics •  Analytics-as-a-Service approach •  Big Data support§  Overall Objectives •  Revenue Maximization •  Social Penetration and User Engagement17
  18. 18. Data is Being Stretched INSTANT INTELLIGENCE Intelligent Data §  Analytics Apps and Algorithms §  Discovery, Visibility, Pattern Extraction Big Data Fast Data §  Streaming, real-time and§  Petabytes vs. Unpredictable Gigabytes §  Mobile app proliferation§  Democratize Analytics §  Non-structured data Flexible Data §  Developer productivity 18
  19. 19. Summary INSTANT INTELLIGENCE§  Cloudification of Analytics is happening§  Predictive Analytics follow the Big Data and Social Web App trends§  Direct Impact in Monetization from Predictive Modeling and Analytics19