Can data science be put in a box?

Can data science be put in a box?

Summary: Data science is one part analysis and one part art. It gathers together data from many sources and gleans important insights. Can this demanding practice be packaged as a piece of software? Prelert believes that it can.

SHARE:
TOPICS: Big Data
1

Prelert's Kevin Conklin, VP of Marketing, and I spent quite some time talking about what his company had done and is now doing to build machine intelligence and predictive analytics into packaged monitoring, management and Big Data. He describes this effort as "putting data science into a box." He means making a sophisticated tool that can review data from many sources, discover the relationships there and predict outcomes rapidly enough to be useful in a real-time computing environment.

What is data science?

Data science can be seen as a combination of business and technical analysis, computer science, applications, modeling, statistics, analytics techniques and math. In moving this study from the academic to the practical, the distinguishing feature is that it must produce actionable, useful results which can help guide an organization to increased efficiency, lower costs and serving its clients better.

It's been said that the data scientist — the practitioner of data science — has to combine computer science, business analysis and artist. 

Conklin believes that the combined use of machine intelligence, pattern matching and sophisticated analysis techniques can package important elements of this science and make it available to organizations and developers alike.

What does data science actually offer?

Organizations have been knowingly — and unknowingly — collecting a massive and growing amount of data about the operations of their business units, their IT infrastructure, and customer behavior.

Sometimes, this collection was part of business operational systems. Other times, this data is hidden deeply inside of the operational logs of systems, operating systems, database engines, application frameworks, applications, storage servers, network infrastructure and even in smartphones and tablets.

Some of this data is structured into relational and non-relational databases. Some of this data is contained in documents, presentations, spreadsheets and — until the development of enterprise searching tools — was difficult to analyze in any automated, systematic fashion.

Prelert, Conklin points out, has been offering this type of capability to its customers.

What has Prelert being doing?

Prelert has developed and is offering tools that make it possible for systems to gather operational data from today's complex IT environment — this includes both physical and virtual computing environments, systems that are housed in the organization's own data center and facilities offered by cloud service providers. These tools automatically learn and then offer actionable insights needed to "detect and resolve developing performance and security issues."

Prelert's goal is offering useful insights in a matter of a few minutes from the time their products are first installed. Conklin said this was like putting a data scientist in a box and making the capabilities available to organizations that don't have the staff expertise to be data scientists.

Where does the company think technology is headed next?

Conklin points out that his company is proud of the technology it has created and believes that organizations and developers could find many other ways to put this capability to work, if only it was packaged in a way to make its use simple and easy. Although he wasn't making any promises about when this would be delivered, he said that Prelert's engineers are hard at work on this project.

Is Prelert alone in having this vision?

Prelert's competition, including companies such as ExtraHop, IBM, Netuitive, and, Wipro, appear to be headed in the same direction. Each believes that its technology should be the foundation of an organization's efforts to apply data science to its business and IT operations.

Will the industry reach a place in which, as my colleague Joe McKendrick says, a Ph. D. in data science will no longer be required (see "Predictive analytics: Ph. D no longer required)? It is hard to say, but Prelert and others are doing their best to get us there.

Topic: Big Data

About

Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. In his spare time, he's also the managing partner of Lux Sonus LLC, an investment firm.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

1 comment
Log in or register to join the discussion
  • Narrow application

    Having been part of an effort to create an "analyst in a box" some years ago, I appreciate the difficulties. The approach of targeting a specific application where the domain knowledge is circumscribed is a good one.

    There is still the risk associated with handing anyone a Magic Black Box that produces answers that no one in the place can reproduce because nobody understands how the answer was formed.
    Robert Hahn