Tagged: Infrastructure Toggle Comment Threads | Keyboard Shortcuts

  • admin 9:53 am on June 22, 2017 Permalink
    Tags: , Infrastructure, ,   

    Building the Machine Learning Infrastructure 

    Latest imported feed items on Analytics Matters

  • admin 9:53 am on October 13, 2015 Permalink
    Tags: , Infrastructure,   

    A Vision for an Analytic Infrastructure 

    by Dan Woods

    An analytic infrastructure can be much like Mark Twain’s definition of a classic: “A book which people praise and don’t read.” An analytics platform is often referred to but rarely architected. Business analysts and data scientists often talk about the power of analytics without talking about the end game. But to make any progress in the big data world – and remain competitive – companies must change the way they think about analytics and implement an analytic infrastructure.

    The current approach to big data analytics is simply unsustainable. For each business question that arises, IT builds a custom application. This application centric approach results in many silos modeled after the operational source. Users can get answers against that silo’s set of data, but they can’t get answers from data across multiple platforms. As a result, data must be constantly moved in and out of the applications, and each application must be maintained.

    It behooves you to think about what you want to achieve with analytics. Most companies today want to become data-driven organizations. In order to do so, however, analytics must be scalable and sustainable so that every department has access to the information it needs to make decisions based on data. An application centric approach is neither scalable nor sustainable. So how can analytics be made more productive for everyone involved and enabled to scale across the entire organization? Instead of hardwiring analytics into an application, you need to find a way to:

    • Apply the right analytic to the right data integration type. Instead of building an application that is essentially a black box, we need a platform that can reach out for all the needed data and then apply the analytic. This approach minimizes data movement and data duplication.
    • Leverage multiple analytic techniques to get insights. You need to build applications in which data is loosely coupled, thereby creating just enough structure to answer frequently asked questions while expanding access to analytics across the organization.
    • Provide self-service analytics for all skill levels. R programming shouldn’t be a requirement for performing analytics. You need an analytics platform that supports a spectrum of users, from data scientists to business analysts.

    The key to enabling these objectives is to make data reusable so that it can be available to as many analytics processes as possible. That means proactively thinking about whether a piece of data will be needed to answer more than one question in the future and understanding where you’re at in your big data journey. You can’t assume that all data will go into a tightly controlled model, like a data warehouse. If you model data using different types of integration based on what you understand about that data, you can create a foundation so that next time you need to answer a question with that data, you can more easily create it.

    In the past, companies had a tendency to over-model and over-integrate their data. Not only was this a waste of money, but it led to an architecture that was difficult to change. Today, companies have the opposite problem: under-modeling and under-integrating. This increases both costs and complexity. A better approach is to invest in tightly coupled integration for high-value data that will be used at scale. Keep other data, of varying levels of maturity, either loosely coupled or non-coupled.

    Taking this approach to building an analytic infrastructure will help you:

    • Meet new needs faster. As the infrastructure grows, the “nervous system” will become more powerful and more easily adapted to meet new needs.
    • Decrease the cost and complexity of the infrastructure. Avoiding application centric silos will reduce the cost and complexity of analytics.
    • Increase productivity. By investing time upfront to make analytics easier for various skill sets, more people can benefit. In addition, new data can be integrated at a lower cost since time and money are not being wasted over-modeling data that will not be used.

    This is a new way of thinking about data, and it may be foreign to many companies. But it is a solid vision for something rarely spoken about but which is necessary for becoming a data-driven organization – an analytic infrastructure. If you’re serious about analytics, then it’s worth working with an experienced advisor who can help you make such an infrastructure a reality.

    Dan-Woods Data Points TeradataDan Woods is CTO and founder of CITO Research. He has written more than 20 books about the strategic intersection of business and technology. Dan writes about data science, cloud computing, mobility, and IT management in articles, books, and blogs, as well as in his popular column on Forbes.com.


    The post A Vision for an Analytic Infrastructure appeared first on Data Points.

    Teradata Blogs Feed

Compose new post
Next post/Next comment
Previous post/Previous comment
Show/Hide comments
Go to top
Go to login
Show/Hide help
shift + esc