Big data analytics enables you to answer a new generation of questions to become an agile business and allows the BI and analytics team to deliver more insights to your organization.
Almost every organization has a large investment in business intelligence. It has developed a tremendous amount of experience in using BI to answer critical questions around “what is happening” in the business. This helps answer everyday questions such as what happened yesterday or last week or macro questions such as what trends happened over the past months or quarters. Analysts can then “slice and dice” this data to see more of what happened.
This paper will examine the style and type of analysis used in big data analytics and show how they are applied to answer a new generation of business questions.
Why? How? When? Where? What happened?
On-the-fly modeling. Virtual columns. Powerful, Easy to Apply Analytic Functions.
Text Analytics. Time-Series. Clustering. Relationship Analytics. Path and Graph Analytics. Recommendations.
Rather than answer what happened, as traditional analytics do, big data analytics answer the how and why and dig deeper into when and where.
On-the-fly modeling – With this capability, sometimes called “schema on read”, the big data analytic platform will build a schema on-the-fly for the specific data used in an analysis without modifying, replacing, or replicating the underlying data. This makes it easy for analysts to create and adjust analytic models almost instantaneously, feeding the iterative process.
Virtual columns -We’ve seen how physical columns in traditional analytics limit the analysis because of the process needed to modify the schema and load the new data. The big data system must allow the analyst to create almost limitless new virtual columns on the fly, applying calculations and/or analytic functions to the data.
Powerful, Easy to Apply Analytic Functions – The big data analytic platform needs to help the analyst run complex analysis on the data through the simple point-and-click application of powerful functions. These are functions that can easily aggregate data, examine distributions, perform text mining, find patterns, identify paths, show clusters, or apply time-series analyses, for example.
Big data analytics complements the analysis performed with traditional tools to dig into these details.
The languages at the heart of traditional BI and analytics – SQL and MDX – struggle with performing the type of analysis to answer big data analytic questions. In these cases, analysts have to go through the time-consuming process of coding their analysis using a combination of a language and analytic libraries such as Python and R.
To meet the goal of being self-service, a big data analytic platform should allow analysts to apply these more complex analytic functions directly to the data via an easy-to-use interface. Let’s look at specific types of big data analysis, examine what business problems they help solve, and how the analysis can be applied easily to the data.