We can reasonably break down the core imperatives of data analytics into three essential areas. Firms that wish to gain the full potential out of engaging in data analytics today are obliged to at least recognize these rudimentary cornerstones as the central building blocks of data intelligence:

1. Prepare
2. Model
3. Automate

By way of clarification, there are probably a whole lot more than three steps to consider here - but calling out this threesome is a prudent step in what (for many IT shops) are still early days with technologies such as Hadoop and its wider world of Big Data.

Data preparation
It's a simple statement to make, but before we can start to work with data it has to be prepared. Before we get to the insightful insights involving dashboards, we need to subject data to combined stages of both machine and human processing.

If we have to define this action, we can say that data preparation is a procedure for gathering, cleaning, amalgamating and consolidating our target information sources into one location (a file, database or table). This process also covers the error correction, merging of data sources, deduplication of extraneous data and accommodating for areas where nulls and zeros exist (or where the data is incomplete in some other way) so that we can progress to building our data model.

Data preparation may also incorporate ‘data enrichment,' i.e., the augmentation of one core set of data with additional intelligence sourced from some other disconnected data stream.

For want of an example here, preparing a movie database would include names of movies with actors' names and movie names and years. But it could also include the interrelationship between actors who have been in the same movie together, or a group of movies that have all been made in the same year and so on. Enrichment could come from identifying background information on actors, directors and more.

Data modeling
Given the myriad and often unanticipated ways data can be consumed, it can be risky to normalize (or integrate) data too far in advance as part of the preparation process... so as we come out of stage one, we enter our data modeling phase.

Data modeling gives us the ability to decide how various elements of data relate to each other and how the heart of the data will behave inside our chosen database. Put simply, data modeling is the classification, documentation and formalization of procedures and events involved within the software in hand. Potentially hugely complex as an overall task, data modeling tools help capture and translate multifaceted system designs into representations of data that are more easily comprehended.

For a retail firm approaching its first major implementation of Big Data analytics, the ability to model the data form allows the business function to specify and describe all the myriad components of the business into data. The model here would include price, product, size, shape, product type, product variants and perhaps also special offers, i.e., the elements of the retail data universe are large.

As complex as data modeling is, the industry now provides drag-and-drop functionality in a variety of products and services to make the entire process easier to manage.

Automate
If you ask a technology evangelist to name the top ten trends for the next half-decade of computing, then ‘automation' will be one of the terms highlighted. Today we understand automation to be the elements within a software data system that we ‘could' build from scratch, but (and here's the crucial point) these are elements that we can more productively find in tools provided by specialists in this layer of computing.

Take the construction of a marketing software system (for example) as a productive and beneficial place to bring in automation software. Automation controls here could include process-driven campaign workflows, built in Business Intelligence (BI), campaign sharing and reuse options, integrated checklists and elements relating to design and navigation. When subject to the positive force of automation, these tasks become more efficient from front to back.

What this brief tour of Big Data shows us, if anything, is that data analytics is no plug-and-play process. There are multifarious multifaceted stages and steps to be aware of. But if we approach this task studiously and in a Zen-like one step at a time fashion, then we can make analytics work for us and our business.

It's as simple as one, two, three - well, kind of.

This post is brought to you by the Big Data Forum, sponsored by SAS.

SAS is a leader in business analytics software and services and the largest independent vendor in the business intelligence market.