Data Architecture for Machine Learning
0 likes209 views
The document provides an overview of McCormick & Company, including: - It was founded in 1889 and has over 11,000 employees worldwide and $4.8 billion in sales in 2017. - Its brands are leading and iconic globally, and its vision is to bring the joy of flavor to life. - It discusses approaches to data architecture including data exchanges, security, and benefits of its technology for self-service and data sharing.
Data Architecture for Machine Learning
- 1. Data Architecture for Machine Learning
- 2. McCormick company profile Founded in 1889 >11,000 Employees worldwide Facilities in 27 countries $4.8 billion in sales in 2017 58% U.S. / 42% International Joint ventures in Mexico and other countries 62% Consumer 38% McCormick Flavor Solutions
- 4. 4 Our brands are leading and iconic across the world
- 5. OUR VISION IS TO BRING THE JOY OF FLAVOR TO LIFE OUR MISSION IS TO MAKE EVERY MEAL AND MOMENT BETTER Passion For Flavor Power of People Taste you trust Driven to Innovate Purpose Led Performance 5 5 Five PASSION FOR FLAVOR POWER OF PEOPLE TASTE YOU TRUST PURPOSE LED PERFORMANCE DRIVEN TO INNOVATEDRIVEN TO INNOVATE
- 6. IBM BLOCKCHAIN IBM Announces Major Blockchain Collaboration with Dole, Driscoll’s, Golden State Foods, Kroger, McCormick and Company, McLane Company, Nestlé, Tyson Foods, Unilever and Walmart to Address Food Safety Worldwide McCormick - Innovation
- 7. COMPUTATIONAL CREATIVITY Ground-breaking technology utilizing artificial intelligence to deliver CONSUMER- PREFERRED- faster, better with even higher consumer preference scores McCormick - Innovation
- 8. Data Services (Data Virtualization) API Management and Runtime Semantics & Discovery Governance Security Technology System 1 System n External API $ Governance Security
- 9. Approach 1. Information is requested on a schedule 2. Information gets external ids, validated and then stored in the Application Database 3. Information must be indexed in order to find it during a search Application Backend Systems External Systems 1 Request Enterprise Data Services 2 Collect Index 5 3 Receive Validate and store 4 ✓Information is directly in application • Inconsistent update times • System outages • Inconsistent information • Coordination challenges Benefits Challenges Data Exchanges – Synchronization Traditional
- 10. Approach 1. Send Search request to the full-text index in the McCormick provided Data Services 2. Receive results Application Backend Systems External Systems 1 Request Search Enterprise Data Services Periodically Index 2 Receive Results ✓Timely Information ✓No replication of information ✓No need to validate information ✓Consistent searching • Index is periodic • Single point of failure Benefits Challenges Data Exchanges – Minimal Synchronization
- 11. Approach 1. Model requests Specific Modifications/Full Information 2. Model incrementally or fully trains Algorithms Backend Systems External Systems 1 Request Enterprise Data Services 2 Collect train 4 3 Receive Benefits Challenges ✓Timely Information ✓No replication ✓No need to validate information ✓Better staging for learning • Potential Data Life-Cycle issues Data Exchanges – Minimal Synchronization
- 12. Data Security Data Security Application Backend Systems External Systems HTTPS (REST, SOAP) Enterprise Data Services *no data is stored here OAuth Services Enterprise Search Index Directory Services HTTPS ODATA JDBC ODBC API Services HTTPS (REST) Data 1 User Authentication to Enterprise Data Services 2 User Authentication to Application 3 Application Authentication & Authorization 5 Role Based Authorization4 Mapping Users to Roles Self-Service – Sharing the Information
- 13. Benefits– Technology ▪ Use same models for self service ▪ Validate service information via HTML ▪ Compare data in multiple systems ▪ Ability to create visualizations on data ▪ Creation of manual spreadsheets based on gaps ▪ Determine quality of proposed data