Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

How AI Will Take Predictive Analytics to the Next Level

DZone's Guide to

How AI Will Take Predictive Analytics to the Next Level

Find out how artificial intelligence will take predictive analytics to the next level and learn about market analysis.

· AI Zone ·
Free Resource

EdgeVerve’s Business Applications built on AI platform Infosys Nia™ enables your enterprise to manage specific business areas and make the move from a deterministic to cognitive approach.

This article is featured in the new DZone Guide to Artificial Intelligence: Automating Decision-Making. Get your free copy for more insightful articles, industry statistics, and more!

Since February of 2018, scientists from Google’s health-tech subsidiary have pioneered innovative ways of creating revolutionary healthcare insights through artificial intelligence prediction algorithms. Based on the back of a patient's eye scan, their system can make predictions against the patient's risk of experiencing a severe cardiac incident.

To achieve this, they trained a Machine Learning system with medical data including the age, blood pressure, and smoking habits of about 300,000 patients. Technologies including machine learning and advanced algorithms can now help data scientists see health issues ahead of time.

Far from Philip K. Dick's Minority Report where humanity's quest was to rule out bad behavior before it happened with guidance from the miraculous precogs, new AI-powered predictive analytics demonstrates a pragmatic purpose for enterprises to execute on.

The evolution of data analysis since the 1980s is quite amazing. In this early age of data intelligence, we were mainly asking, "What happened?" To answer this question, IT teams were building statistics and more or less interactive reporting. Then, in the 1990s, we started to talk about analysis based on MS Excel. OLAP (online analytical processing) appeared at the same time. The analysis era aimed to answer the question, "Why did this happen?" Then, in the 2000s, companies began to specifically focus on, "What's happening now?" The advent of dashboards and scorecards built the monitoring generation, which later gave rise to a new question: "What will happen?" To answer this, data analysts first used advanced statistics, data mining, and advanced data analytics. Technologies including machine learning, neural networks, and deep learning now help data scientists face the challenges raised by the need for prediction.

What Is Predictive Analytics?

Advanced analytics solutions consist of a comprehensive combination of elaborated methodologies, technologies, and infrastructures that take analytic processes over and beyond traditional data processing. Its purpose ranges from making predictions to bringing to light actionable insights.

Advanced analytics' most famous application scope is big data analytics, which aims to unveil patterns, subtle correlations, and trends to empower decision-making processes.

At a higher level, when leveraging data mining methods, complex statistical models, and machine learning technologies, advanced analytics allows for making effective data-based decisions and building sentiment analysis or recommendation systems that lead to predictive analytics.

Predictive analytics should consider these four axes: prediction, speed, business, and accessibility.

  1. Prediction: Predictive analytics goes beyond the standards that allow for producing simple descriptions. In a few words, descriptions let decision-makers understand the current state of their business, while predictions empower them to implement action plans knowing what may or may not happen next.
  2. Speed: The strength of prediction capabilities also lies in the ability to come up with actionable outcomes quickly, compared to a business intelligence batch that could require a whole night, and sometimes even days, to see calculations run their course to the end.
  3. Business: These analytics are business-oriented by their very nature, far from statistical research where you just search for trends within a dataset about past activities.
  4. Accessibility: The combination of the three previous axes tends to strengthen the need for business accessibility. The link between the prediction solution and its outcomes, and the decision-makers who will put it into action, requires it to be as simple and as natural as possible.
"Information is the oil of the 21st century, and analytics is the combustion engine." — Peter Sondergaard, Gartner

Market Analysis and Importance

Forrester forecasts a 15% compound annual growth rate for the predictive analytics market through 2021. They observed that within a very large inventory of AI-related innovations, including algorithms and solutions, a surprisingly large portion lies in the open-source community.

In their 2018 Magic Quadrant for Data Science and Machine LearningPlatforms, Gartner reveals that on top of historic big players, traditional software editors are shifting from classic descriptive and diagnostic analytics to predictive and prescriptive analytics. Far from the mainstream, advanced analytics are no longer the preserve of Google, Apple, Facebook, and Amazon (GAFA).

Studying the worldwide advanced and predictive analytics software market, IDC reports the trends and momentous evolutions within the analytics market. They state that we are reaching a point where companies have a perfect knowledge that, besides the still relevant value of business intelligence tools, they now have to derive the greatest benefit from the value of "forward-looking analytics" — AKA predictive analytics.

"Prediction is very difficult, especially about the future."—Niels Henrik David Bohr (Nobel Prize for Physics in 1922)

The Scope of Predictive Analytics

Each and every modern business domain can gain maximum potential from predictive analytics. In all of them, what is key is that they are in possession of sufficiently large, various, detailed, and reliable historical data that are enriched with the latest types of "intelligent" data, including those coming from smartphones, connected devices, sensors, logs, etc. All types of companies must be determined and consistent in their desire to leverage their data in optimizing their processes. This applies equally to fraud detection, processes optimization, costs reductions, market trends anticipation, and the discovery and innovation of new business opportunities, for example.

Let's go through four examples.

1. Fintech and Banking

Regarding the use of a credit card, abnormal behaviors can be revealed by establishing models and observing specific patterns of inappropriate use. This is known as fraud detection, and can also apply to other industries. Social media abounds with data that financial companies are exploiting in order to better understand, get closer to, better satisfy, and ultimately retain their customers. It provides them with extremely valuable data that can help them predict the behavior of the market and their customers.

2. Oil and Gas

This is a domain that takes a huge advantage of the Internet of Things (IoT) — more precisely, IIoT (Industrial IoT). For example, GeneralElectric is using data from vibration and thermography systems and also additional extraneous information to extrapolate data, establish correlations, and make predictions in a marine exploration or extraction site. To explore how AI and IoT synergize, refer to AI and IoT: Taking Data Insight to Action.

3. Retail

Prediction systems are quite commonly observed in the retail industry to improve engagement and personalization for consumers, as well as delivery processes or stock management. Combined with marketing research and actions, these systems could also aim to anticipate trends or evaluate the adoption rate of a new brand or service.

4. Industry

This business domain fully takes advantage of its data, even more so from its IIoT data, implementing predictive maintenance. The smart devices, sensors, and intelligent devices they operate produce tons of data and logs that they leverage through models and advanced algorithms. This is critical for them, as it allows them to optimize their production and obviously reduce risks and cost. But the main point is that IIoT devices allow them to better understand how their systems work and can be maintained — whether they are making industrial machinery, airliners, trucks, or wind turbines.

AI Takes Predictive Analytics to the Next Level

Forrester expresses their conviction that companies that want to leverage AI should start with a predictive analytics machine learning system. They state that machine learning is fundamental to artificial intelligence. Let's see how artificial intelligence can enhance predictive analytics.

Healthcare

Machine learning models using echocardiographic data can greatly improve mortality predictions. The amount of data that physicians have at their disposal is so substantial that they don't have material time to extract their maximum potential.

The Geisinger Health data science team in Danville, Pennsylvania worked on data from 170,000 patients for a total of 330,000 Doppler ultrasonography results. Machine learning models and related algorithms let them significantly improve their search field. It pointed out that while 500 measurements derived from echocardiography, six echocardiographic, and four clinical variables were the most important for the five-year estimates.

Retail

In the worldwide retail market, Alibaba Group is a real phenomenon. On just one day (Nov. 11, 2017), they made more than $25 billion, which represents roughly 26x that day on Amazon Prime. The strongest activity in terms of payment at this time was 256,000 payments per second.

Alibaba is candidly crediting their success to artificial intelligence.

The Alibaba data science lab has crafted a "system which uses real-time online data to predict consumer wants, and the models are constantly updated for each individual through AI to take into account purchase history, browsing history, and online activities. "They called it "E-Commerce Brain."

I-ERP

Enterprise Resource Planning (ERP) software solutions provide organizations with the ability to manage their business. They integrate and centralize their business data, processes, and workflows across different departments in the enterprise to offer 360-degree views of activities. It includes accounting management, financial, manufacturing, production, sales and distribution, human resources, customer relationships, and more.

Combined with AI technologies, i-ERPs (Intelligent ERPs) are designed to track, route, and analyze these processes, and make predictions. They handle repeatable tasks and utilize voice recognition, machine vision, and natural language processing to interact with humans and assist them.

Conclusion

Predictive analytics takes full advantage of enormous technological breakthroughs of the data intelligence era including big data analytics, Internet of Things, cloud, and artificial intelligence. Companies are now aware of the richness of their data and intend to derive the maximum benefit from it to run their business.

Artificial intelligence is grouping together several technological innovations including speech recognition, virtual assistants (AKA bots), machine learning, deep learning, machine vision, biometrics, robotic process automation (RPA), text analytics, and natural language processing. There's no doubt that predictive analytics has gotten stronger and is gaining more credibility in the land of digital transformation.

This article is featured in the new DZone Guide to Artificial Intelligence: Automating Decision-Making. Get your free copy for more insightful articles, industry statistics, and more!

Adopting a digital strategy is just the beginning. For enterprise-wide digital transformation to truly take effect, you need an infrastructure that’s #BuiltOnAI. Click here to learn more.

Topics:
artificial intelligence ,machine learning ,predictive analytics ,market analysis ,healthcare ,retail ,i-erp

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}