A Brief Primer on Data Analytics

As one might expect, the SAS Global Forum 2013 in San Francisco is focused heavily on the benefits of data analytics. The problem with describing analytics is that it’s either spoken of too broadly (“It’s important to analyze data”) or in a way that’s so technical, it’s almost as if the vocabulary was developed by aliens. Here’s a brief primer on how data analytics works and what it is.

How does data go from collection to analysis?

To have data analytics, businesses first need a data collection mechanism (which grabs information coming from different sources like point-of-sale devices, social media, inventory management, and customer contact information) and a warehouse—something provided by vendors like Teradata or Hadoop. Once the data is collected, it needs to be validated and cleansed.

Following this process, the data must be staged—essentially placed into a container ready for analysis. The analytics tool comes in, assess the data from different perspectives, and spits out hopefully relevant results.

What are the capabilities of analytics?

The capabilities of analytics depends greatly on what type of tool marketers are using. Jim Davis, SVP and CMO of SAS, places these capabilities on a spectrum. At the lowest end, he describes summary statistics—basically any report that sums up the content of a spreadsheet column. For instance, identifying the top 10 best-selling items on a list is an analytics application. Microsoft Excel has this ability, so it’s pretty rudimentary.

The next stage is to identify trends over a historical period of time. Typically, this is when users require a better interface as data management becomes more complex.

The third stage—and the one that Davis feels differentiates analytics providers—is the ability to anticipate future trends and situations. For instance, a bank might run regression analysis to determine the risk of a certain investment. In other words: It might look at how various factors relate to each other (What is the relationship, if any, between really bad weather in Taiwan and a slowdown in semiconductor manufacturing?) to determine what will likely happen next.

The fourth stage is the incorporation of operations research techniques. This means using advanced and predictive analytics to make optimal business decisions. For instance, if a retailer has a certain amount of products in-store, based on historical data, it can decide how best to price the product for the next week.  

What is in-memory analytics?

The term “in-memory analytics” has been bandied about lately, though the concept isn’t new; the interest in Big Data tools has simply raised its profile.

In-memory analytics enables businesses to analyze and make predictions on volumes of data extremely quickly. This is because data usually needs to be moved from its storage area and transformed into a format that the analytics engine can read. For large datasets, this can take days. In-memory analytics skips these lengthy stages, because the analytics tool can read and interpret the data directly from the database, without having to export or format it.

To analogize, imagine an English speaker having to analyze a brochure written in Mandarin Chinese. He’ll have to take the brochure to a translator who will herself have to read it before translating it into English and sending it back to the English speaker.

With in-memory analytics, imagine the English speaker magically learns how to speak Mandarin. He can read and interpret the brochure on his own, expediting the process.

Related Posts