Gartner Data & Analytics Summit
March 6–9, 2017 at Grapevine, Texas, United States
Solution Provider Session Report
Predictive Analysis Automation: Analytics at the Speed of Business
Ryohei Fujimaki, Ph.D., Research Fellow at NEC Corporation & Mr. Tomohiro Oka, Vice President of IT Innovation Department at Sumitomo Mitsui Banking Corporation
New predictive analysis automation technology set to transform data science and business operations
Today, everyone is talking about predictive analytics, and many businesses have started using it for a variety of purposes, from optimizing advertising campaigns to attract new customers, to optimizing ordering and inventory through accurate supply and demand, or optimizing maintenance schedules to avoid system failures. Without a doubt, predictive analytics has a huge potential to accelerate your business. However, current predictive analytics technology has significant limitations. It requires highly talented data scientists and a considerable amount of time to deliver accurate prediction models.
NEC is offering a new game-changing artificial intelligence (AI) technology, predictive analysis automation, which disrupts predictive analytics, and ultimately changes the way businesses operate. NEC’s predictive analysis automation technology generates highly accurate and transparent predictions in just hours. Automation enables anyone from data engineers to systems engineers and business analysts to produce as good prediction models as a talented data scientist with a Ph.D. in computer science. This will not only help alleviate the shortage of data science expertise, but can also help talented data scientists improve the productivity of their predictive analytics as much as 100 times.
Self-service analytics is a new and increasingly attractive concept for many enterprises, so for this comparative exercise, let’s focus here on a new page of self-service analytics: self-service predictive analytics. What are the challenges that customers faces with traditional self-service predictive analytics, and how does NEC’s predictive analysis automation address those obstacles, and facilitate the accurate data analysis to support buoyant business growth?
How does predictive analytics work?
Business goal setting: The most important step of predictive analytics, which defines the right business goals, what they are going to predict, and how to apply the model to create additional revenue or reduce costs.
Data preparation and feature engineering: The most time-consuming step which creates a single data table for machine learning from hundreds of heterogeneous data sources. This represents approximately 80% of the analytics process and is the hardest part of data science.
Predictive modeling: Once a data table is prepared, a prediction model is built by choosing the best algorithms and spending a vast amount of effort tuning parameters.
Prediction: Deploy the prediction, and obtain prediction result by simply running the learned model.