Source: IBM developerWorks
Big data classification is not a new concept. It’s been around for a while, people just didn’t realize it for the most part. What changed is that now companies know what it is and use the results to find new clients. The question then becomes, how is the architecture created to handle all the different information out there? First, you have to classify the data, then look for the right architecture that you’ll need to use. Classifications are actually big data problems that your company has. That’s why you need to classify the data first, then decide on what kind of architecture you’re going to need to fix the problem.
Big Data Classifications
Utilities: Predicting Power Consumption – Data creation completed by machines. Uses smart meters to measure consumption and power grids. Big Data solutions need the ability to analyze supply and demand.
Telecommunications: Customer Churn – Data creation done by reviewing social media, Web and transaction data. You need to create detailed churn models to keep up with the competition. Big Data solutions can help by using predictive analytics.
Marketing: Sentiment Analysis – Completing data gathering through using social media and the Web. Sentiment and profile data needs integration to find any useful results.
Customer Service: Call Monitoring – Human generation creates the data needed here. IT departments need the ability to analyze application logs in standardized formats to create Big Data files.
FSS/Healthcare: Detecting Fraud – Creating data files here is through using machine, transaction and human generation. This needs real-time or near real-time monitoring to be effective. There’s no other way to react quickly if there are reporting of unusual activity.
Make sure that when you’re classifying the data that you look for characteristics. This will help you to figure out what kind of architecture you’re going to need to create a Big Data platform for your agency.
Big Data Architectures
Analysis Type – Is analysis completed real-time or saved for later? Consequently, choices here will decide types of tools, hardware and data sources to name a few. Because of this, it can affect how you want to analyze the data.
Choosing a Processing Methodology – This is where you’ll choose the techniques that will be used to process the data. Your business requirements will decide which technique is going to be used. Choose wisely. It can be either predictive or analytical models, ad-hoc query or report building. Which do you need more for your business?
Data Frequency and Size – Knowing these two things will help you to decide which storage unit, formats, and tools you need. Consequently, the frequency and size depend on data types also.
The types of data that need gathering, plus content formats which are key to choosing which tools and techniques are going to be used are extremely important. Considering what the source is of the data is important too.
Data Consumers –
People in different business roles
Different data repositories and/or applications
Hardware – Make sure to choose the right hardware. Because you understand limitations of any hardware that you decide to buy to support this, it so helps you decide on the solution for the company.
A lot goes into choosing which big data architecture is best for your company. Don’t rush into it, take your time when figuring this out as it will cost money. Think about which area you need the most help with, then decide on which architecture you’ll use. It’ll help in the long run to save money and gain more customers.