The Big Deal With Big Data

As machine learning and predictive analytics become more sophisticated, companies can base decisions on evidence, and deep learning will push the boundaries even more with better problem-solving and language comprehension. Are you ready?

Experts predict that the universe of data—or 'dataverse'—will reach 180 zettabytes by 2025. It's a truly mind-boggling number, highlighting the exponential growth of big data. Bernard Marr, author of Data Strategy: How to Profit from a World of Big Data, Analytics and The Internet of Things, offers some perspective, noting that 90 percent of existing data in the world has been generated in the last two years.

Unfortunately, the volume and variety of data available does not always equate to value.

How can companies harness big data effectively? Harvard University's Gary King suggests that "Big data is not about the data!" Instead, he writes, "Although the increase in the quantity and diversity of data is breathtaking, data alone does not a Big Data revolution make. The progress in analytics making data actionable over the last few decades is also essential." That's where Artificial Intelligence (AI) comes in.

Moving Artificial Intelligence Out of the Sci-Fi Realm

AI is often misunderstood, rooted in Hollywood stereotypes of rogue robots and manipulative mainframes. But the reality of AI is far less villainous. AI can be broken down into two main types:

• Narrow—Also known as Weak AI, Narrow AI leverages big data for very particular tasks. Predictive analytics, for example, can be used to identify patterns and correlations in data—from forecasting the weather to analysing news data to anticipate risk.
• General—Also known as Strong AI, General AI involves human-like cognitive abilities. Still, there are limits to these cognitive skills. Machine learning falls within this type; while it doesn't enable abstract 'thought,' it does support adaptation and continuous improvement.

Some consider Artificial Super Intelligence a third type, but right now, it is more theoretical than actual. IBM's Watson beating Jeopardy champions or Amazon's Alexa responding to requests may mimic human responses, but as yet, computers aren't making true cognitive leaps. Instead, they simply access a wealth of big data far faster than any human can. For now, the Terminator remains a fiction—but companies that fail to embrace AI risk terminating their existence as competitors who are faster to adopt—and extract value—from data implementations.

Netflix is a prime example of a data-driven leader. As the popularity of streaming video climbed—and video rentals tanked—Netflix embarked on a remarkable journey of using data analytics to grow its user base, enhance customer retention and inform programming decisions. The company says that its algorithms save $1 billion annually in customer retention. In addition, analytics have inspired several critically acclaimed and fan favorite series produced by Netflix including House of Cards, Stranger Things and Orange is the New Black.

Overcoming the Challenges of Extracting Value from Big Data

While most companies recognize the importance of implementing big data initiatives, many still struggle with the execution. Many challenges are organizational in nature—from attracting and retaining data specialists to breaking down organizational data silos to make better use of internal datasets. Nearly 70 percent of companies have made establishing a data-driven culture a priority, according to a 2017 Harvard Business Review article, but only 40 percent are hitting the goal.

Where should you start? Start by deciding what business-critical questions you want to answer. Then look at what types of data you need to uncover the answer. Some of the data may be available internally; some may require some effort to extract from various departmental silos; some may be available in the public domain; and some may be available through Data as a Service (DaaS) providers. Moreover, consider how the insights you uncover will be shared with key stakeholders. Making data accessible means eliminating technical jargon and telling a compelling story. Data visualisations breathe life into raw data, making it easier to digest while showcasing essential points.

With the right people, processes and datasets in place, however, organizations can achieve actionable business intelligence—and measurable ROI—from machine learning, predictive analytics and other data implementations.

• Discover hidden insights—AI can help companies detect patterns ranging from the root causes of customer attrition to emerging trends that can inspire new products or services. Regression modelling using company, industry and economic data can help organizations understand both the real-time effects and long-term consequences of government policies or market changes. Analytics of news and social media data can complement internal customer data to enhance marketing reach and effectiveness.
• Automate business processes—Hedge funds can leverage advanced AI analytics to enable high-frequency trading, from identifying patterns that led to poorly executed trades and 'learning' from those patterns to improve future trade performance to predicting stock market movements based on both historical and current market data. Banks can leverage machine learning algorithms and predictive analytics to automate fraud detection or quickly identify sanctions risk among customers.
• Reduce disruption—A lack of insight is often at the core of disruptive events. Thanks to predictive analytics and Internet of Things (IoT) data, manufacturers can circumvent disruptions on the assembly line by knowing when preventive maintenance is needed to avoid a production slow-down. Likewise, data analytics empower PR and marketing organisations for faster response times when a crisis—or an opportunity—emerges.
• Realise savings—The operational efficiencies supported by big data and analytics can reduce expenses while freeing up your human resources for the work that cannot yet be done by AI. In addition, the ability to optimise decision-making—whether related to buy/sell decisions of a stock trader or launching a new product—opens the door to creating measurable value.

As machine learning and predictive analytics become more sophisticated, companies can base decisions on evidence, and deep learning will push the boundaries even more, with better problem-solving and language comprehension. Are you ready?

Mark Dunn is Director of Nexis Data as a Service, Nexis Solutions
Image Credit: Nexis Solutions

To view the original article, visit the SmartZip blog.