What Is Augmented Analytics?

September 10, 2019 | 5 minute read
Michael Chen
Senior Manager
Text Size 100%:

In recent years, the term augmented analytics has joined the conversation around how data scientists extract insights from big data. Rather than representing something completely new, augmented analytics refers to the convergence of business intelligence and emerging computer science fields such as artificial intelligence (AI) and machine learning (ML). Today, augmented analytics is gaining steam, transitioning from the industry’s next big thing to a must-have tool. As the next evolution of the foundation built by BI, analytics, and big data, augmented analytics combines many emerging technologies for a platform that delivers insights at a previously unheard-of speed and level of accuracy.

In the past decade, business intelligence has focused on gathering data from various sources, then processing and outputting it in dashboards and visualizations. This has brought insight to the business world, enabling data-driven decisions. However, this required significant manual preparation from various departments. With advancements in AI and ML, much of this can become automated—and the result is the next big leap forward in business technology.

How Business Intelligence Works

Business intelligence systems emerged in the 2000s, from standalone database tools existing solely on individual desktops to modern systems that connect with multiple data sources and focus on data manipulation. The key here is the ingestion of data; without data, the whole concept of business intelligence is never able to take off. Business intelligence systems evolved as processing power and network connectivity grew, expanding the scope of capabilities to go from an analyst’s local desktop machine to a larger, more interconnected platform.

Currently, business intelligence systems excel at ingesting big data from multiple sources, allowing analysts to take a deeper look at past activity. This results in numerous business insights that explain how and why things have happened, then provides the tools to create visualizations and charts to tell the story surrounding the data. In turn, this enables data-driven decision making.

How Artificial Intelligence and Machine Learning Work

AI and ML are often used interchangeably or even confused with each other, and that’s not totally accurate. AI refers to the larger umbrella of enabling systems to make decisions like humans—that is, making smart decisions based on context and available information rather than simple if/then programming.

ML is a key component in this; ML examines data and looks for patterns which can then provide the context for AI-based decisions. For example, ML might review the processing of data transactions for a credit card company and quickly look for anomalous patterns in user data. AI then flags this and determines if it is a potentially fraudulent transaction worth investigating by customer service.

How Augmented Analytics Brings Business Intelligence, AI, and ML Together

Business intelligence is all about creating data insights. AI and ML are all about learning from large datasets for machine-driven decisions. AA, then, uses the foundation of business intelligence and then adds ML/AI on top of it. A good way to frame this is by considering the current process involved with using business intelligence. As it stands right now, a business intelligence platform ingests data from multiple sources before IT departments prepare the data and data scientists process it for analysis.

An augmented analytics system takes those latter steps (data preparation and initial analysis) and automates them using ML and AI. A simplified explanation is that machine learning handles the data preparation (processing the ingested data, preparing the relevant data, looking for patterns), and AI handles the initial analysis (using models and algorithms built by data scientists). It’s a little more nuanced than that but that’s a good surface-level way to understand how augmented analytics works. Consider the manual labor used in a traditional system:

  • Data preparation by IT staff involves exporting datasets, then combining, structuring, and organizing them for further analysis. If your dataset includes thousands of records, or millions of records, this could require significant hours of preparation per request.
  • Initial analysis by data scientists can be an intensely manual process involving examination of countless records to look for patterns and dig for insights. Many datasets require a first level of analysis, which takes care of broad-stroke conclusions before diving in deeper; this can be recognized automatically using ML and AI, reserving the data scientist’s bandwidth for more intense work.

With ML and AI working in the background 24/7, this process is constantly active. That means that the ML algorithm is constantly refining patterns while looking for new ones. At the same time, the overall AI model is improving through sheer volume of data; the more data consumed, the more accurate the model. This automation streamlines processes, removing manual steps to drill down to relevant data faster. In addition, natural language processing (NLP)—the same technology that powers AI virtual assistants such as Siri and Alexa—means that tasks shift from data preparation to discovery.

Benefits of Augmented Analytics

Augmented analytics provides many of the same benefits as business intelligence, but also delivers a level of efficiency and accuracy only available via computer processing. Thus, the true scope of augmented analytics goes past business intelligence’s native capabilities, including:

Increased accuracy: When data scientists manipulate multiple datasets to prepare for analysis, it is statistically likely that a mistake will occur during that process. The larger the volume of data, the greater the possibility of a mistake, and the longer it takes to run checks for mistakes. When utilizing machine learning for these types of processes, such mistakes are eliminated.

Increased speed: There are two process gaps that can occur when initiating a project with standard business intelligence platforms: the time required to manually prepare data as well as the wait time for associated parties to respond to requests. With AA, request processing begins immediately once the request is submitted, launching the internal AI to cull the appropriate data and begin drilling down to the specific output for the project—all at the speed of a machine, not a human.

Reduced bias: The term “bias” often has a negative connotation, but bias doesn’t have to imply a personal shortcoming. Instead, it can simply refer to habits and routine. As human beings, we revert to patterns in process. Thus, there may be a blind spot for a data scientist because of a personal process that unintentionally overlooks one potential aspect. That type of bias, while not malicious, can lead to overlooked insights. In this case, machines will work more thoroughly and more efficiently without this inherent bias.

Increased resources: There’s a common—and unfounded—assumption that any movement towards automation and AI will reduce the responsibilities of IT staff or data scientists. In fact, the exact opposite is true; augmented analytics can actually increase the value of both because it frees them from manual labor so they can focus on more important tasks. For IT staff, that means supporting ever-growing demands on hardware and connectivity, and for data scientists, it means much more time creating deeper insights. In short, everyone wins with augmented analytics.

Learn More About Augmented Analytics

We’re just on the cusp of the augmented analytics revolution for business intelligence. To get ahead of the curve, make sure you download the new ebook What Is Augmented Analytics? And for more about how you can benefit from Oracle Big Data, visit Oracle’s Big Data pageand don't forget to subscribe to the Oracle Big Data blog to get the latest posts sent to your inbox.

Michael Chen

Senior Manager

Previous Post

Migrating Your Big Data to the Cloud

Michael Chen | 2 min read

Next Post

Big Data Highlights from Oracle OpenWorld 2019

Michael Chen | 4 min read