Harmonizing Data for the Warehouse

649 views

Published on

Be certain data scientists can trust their data

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
649
On SlideShare
0
From Embeds
0
Number of Embeds
65
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • As we learned fromArt Krulish on Tuesday, omni-channel marketing presents a set of new challenges for retailers--and their data warehousing professionals.<Click>Today’s highly-connected marketplace offers more choices, which makes understanding customers more important than ever. And customers realize this too. In fact, many are now expecting companies to cater to them like never before.<Click>Smart devices and social media sites have up opened up new ways for customers to interact with retailers. The result is spreading customer interactions across many channels. A key task for data professionals is to figure out which channels contribute most to a reliable view of their customers.<Click>And with these new channels comes lots and lots of data. Some of the data may come in semi-structured forms that are unfamiliar to the data professional. Figuring those out, and the cost of processing large volumes of it, makes it critical for data professionals to be certain they are crunching only the data that will improve customer insights.<Click>Some of this information is fleeting. If I stop in at a coffee shop, I’ll be in the area long enough to log onto WiFi, check-in on Facebook, and drink my latte—then I’ll be off. If I get an offer that night to check out the gym next door, it’s probably too late. Unless I visit that coffee shop several times a week. Detecting patterns in customer behavior separates the noise from real opportunity. <Click>But in order to get to those trends, we need both short and long term views—in short we need an enterprise view of the customer.
  • As we learned fromArt Krulish on Tuesday, omni-channel marketing presents a set of new challenges for retailers--and their data warehousing professionals.<Click>Today’s highly-connected marketplace offers more choices, which makes understanding customers more important than ever. And customers realize this too. In fact, many are now expecting companies to cater to them like never before.<Click>Smart devices and social media sites have up opened up new ways for customers to interact with retailers. The result is spreading customer interactions across many channels. A key task for data professionals is to figure out which channels contribute most to a reliable view of their customers.<Click>And with these new channels comes lots and lots of data. Some of the data may come in semi-structured forms that are unfamiliar to the data professional. Figuring those out, and the cost of processing large volumes of it, makes it critical for data professionals to be certain they are crunching only the data that will improve customer insights.<Click>Some of this information is fleeting. If I stop in at a coffee shop, I’ll be in the area long enough to log onto WiFi, check-in on Facebook, and drink my latte—then I’ll be off. If I get an offer that night to check out the gym next door, it’s probably too late. Unless I visit that coffee shop several times a week. Detecting patterns in customer behavior separates the noise from real opportunity. <Click>But in order to get to those trends, we need both short and long term views—in short we need an enterprise view of the customer.
  • As we learned fromArt Krulish on Tuesday, omni-channel marketing presents a set of new challenges for retailers--and their data warehousing professionals.<Click>Today’s highly-connected marketplace offers more choices, which makes understanding customers more important than ever. And customers realize this too. In fact, many are now expecting companies to cater to them like never before.<Click>Smart devices and social media sites have up opened up new ways for customers to interact with retailers. The result is spreading customer interactions across many channels. A key task for data professionals is to figure out which channels contribute most to a reliable view of their customers.<Click>And with these new channels comes lots and lots of data. Some of the data may come in semi-structured forms that are unfamiliar to the data professional. Figuring those out, and the cost of processing large volumes of it, makes it critical for data professionals to be certain they are crunching only the data that will improve customer insights.<Click>Some of this information is fleeting. If I stop in at a coffee shop, I’ll be in the area long enough to log onto WiFi, check-in on Facebook, and drink my latte—then I’ll be off. If I get an offer that night to check out the gym next door, it’s probably too late. Unless I visit that coffee shop several times a week. Detecting patterns in customer behavior separates the noise from real opportunity. <Click>But in order to get to those trends, we need both short and long term views—in short we need an enterprise view of the customer.
  • The Kalido information Engine is the first highly-automated, purpose-built environment for implementing agile data warehouses.Business modeling allows IT and business to speak the same language as they collaborate because the requirements and business rules are defined graphically and then automated by Kalido.And because a Kalido warehouse is highly automated, things move quicker in development by reducing etl coding, removing the need for unnecessary translations into to logical and physical models, as well as having developers make changes to the physical layer directly.Finally, Kalido delivers information in a variety of formats, including industry standard BI tools like Excel, Qlik View, Business Objects, Cognos and SSIS. Automating the BI semantic layer greatly reduces the time required to build reports, improving your reaction time to customer trends.
  • Let’s take a few moments to describe a business model in a little more detail. This model describes a brick and mortar electronics retailer. As you see, the model represents real world aspects of the business such as the Products, Customers and internal organization. These domains are drawn as tan square boxes. Within them are individual elements that describe these domains in more detail, such as credit rating and product class, which are drawn as light blue square boxes. Together they describe the reference data of the organization and they do so using business terms, not technical ones. For example, the similarities and differences between Corporate and Individual Clients is made clear and graphical. In other warehouse environments, these simple relationships could be fragmented across numerous normalized tables and, therefore not nearly this easy to interpret.The black lines you see describe the relationships and hierarchies of the organization. Solid and dashed lines represent mandatory and optional relationships. The lines that loop back to the same object denote a hierarchy within that object itself.The activities undertaken by the company, most commonly described as transactions, are represented by the rounded, colored boxes. Sales Revenue and sales projections from brick and mortar stores, for example. Lines of the same color as the transaction box connect it the reference data that are directly captured. From there, transactions are easily summarized up any of the reference data hierarchies, making it easy to get to analysis quickly using Kalido.Even as these models grow, they remain relevant to the business. They are the first place those users as well as data scientists look to understand what is possible with the data—and how new possibilities can be added. To show you how the business model drives the solution, let’s quickly demonstrate how this brick and mortar retailer would add it’s first new channel.
  • --Staging Layer Definition--Integration Layer DefinitionNext I’ll show the consumption layer definitionDemo deploying the Corp Demo Model--BIM (deploy)--Explorer (show then build att & map tables)--Open SQLServer and show tables?
  • (TD_DEMO_3)--Walk through the UID screens--Show the result modelFeel like it takes too long to generate the UID but am open to discuss.
  • (TD_DEMO_3)--Walk through the UID screens--Show the result modelFeel like it takes too long to generate the UID but am open to discuss.
  • The Kalido Information Engine offers many advanced capabilities, like:Comprehensive Hierarchy Support, which enables business and technical teams Express complex business relationships simply and graphically/History tracking and Audit, where Kalido maintains history of both the data AND the model. Kalido warehouse operations automation simplifies the process of automating loads, test switches and load dependencies in Kalido and move to scheduling toolThese capabilities make the Kalido Information engine the right warehousing platform for deriving the right analytics from your Omni-ChannelOrganization to fully understand your customers.Thank you for attending today’s event.
  • Harmonizing Data for the Warehouse

    1. 1. © 2013 Kalido I Kalido Confidential I June 5, 20131Harmonizing Data for the WarehouseBe Certain Data Scientists Can Trust Their DataJune 4, 2013
    2. 2. © 2013 Kalido I Kalido Confidential I June 5, 20132AgendaNeeds of the Data ScientistAssessing the dataLoad and integrate new dataStewardship is key to trust in dataMake it available fastDemonstration
    3. 3. © 2013 Kalido I Kalido Confidential I June 5, 20133Opportunities and Challenges of Data ScientistsData scientists need new data fast, but can’t sacrifice accuracy!Forward lookingRequirements are murkyNeed data they can trustFrequently requires new data
    4. 4. © 2013 Kalido I Kalido Confidential I June 5, 20134Assess the Data to Answer the QuestionData Scientists need map to quickly assess the data they have now!What data is available?Is it all the data I need?How do I tie it together?
    5. 5. © 2013 Kalido I Kalido Confidential I June 5, 20135New Data – New InsightsNew insights are the business of the data scientist.What other data improves my insights?Can I get it yesterday?Can you align it with the data I have?Will it be accurate?
    6. 6. © 2013 Kalido I Kalido Confidential I June 5, 20136DataScientistSandboxManageSandboxes• Metadata• Trusted Data• BI SemanticLayersKalidoInformationEngineAutomation• ETL• Business Rules• Physical Layer• Stewardship• Sandbox SourceWarehouseBusinessInformationModelDefineRequirementsGraphically• Data Modeling• Metadata• Business RulesAgile DW: Define Data Science Needs, Shorten data integration time, Delivers New Insight Fast!Real Agile Data Labs and SandboxesWorkflow
    7. 7. © 2013 Kalido I Kalido Confidential I June 5, 20137Business Information ModelingCapturesrequirementsusing businessterms, nottechnical onesClear view ofwarehouseartifactsAdd artifacts bydrawing themThe model drivesthe solution“What sales channel did a customer on-board?”“What are the product mixes by channel?”
    8. 8. © 2013 Kalido I Kalido Confidential I June 5, 20138Kalido Product DemonstrationDemo
    9. 9. © 2013 Kalido I Kalido Confidential I June 5, 20139DemonstrationRefactor the BusinessModel
    10. 10. © 2013 Kalido I Kalido Confidential I June 5, 201310DemonstrationIntegrate & StewardNew DataWorkflow
    11. 11. © 2013 Kalido I Kalido Confidential I June 5, 201311DemonstrationAutomate ResultsGeneration
    12. 12. © 2013 Kalido I Kalido Confidential I June 5, 201312Key Kalido Information Engine CapabilitiesSophisticated ModelingHigh Performance Data Loading and IntegrationData MatchingWorkflowData Authoring InterfaceComprehensive Hierarchy SupportPowerful Results GenerationInexpensive Change ManagementSecurity & AuditHigh Performance Platform Support
    13. 13. © 2013 Kalido I Kalido Confidential I June 5, 201313For More InformationVisit http://get.kalido.com/harmonize to...Check out our end-to-end demonstration seriesTune in for our next Data Scientist Summer Session:Rapid Iteration Methodology Using ModelingGet the Business Information ModelerRequest a demo

    ×