Five successful cases of machine learning !!

Nowadays, artificial intelligence and machine learning are becoming more and more popular in the corporate world. Organizations are increasingly using these technologies to predict customer preferences more accurately and strengthen business operations.

According to the well-known research company IDC, by 2023, the expenditure on artificial intelligence systems will reach 97.9 billion U.S. dollars, almost three times the expenditure of 37.5 billion U.S. dollars in 2019. According to a research released by Capgemini in June this year, 87% of the 950 enterprise organizations surveyed have deployed AI pilot projects or have put limited use cases into practical applications.

However, the outbreak of the new crown epidemic has brought a new challenge to artificial intelligence, because since March, many business organizations that rely on historical data to improve algorithms have found that their models have been biased. Jerry Kurtz, executive vice president of insight and data at Capgemini Consulting, said that this “data drift” phenomenon makes it difficult for companies to rely on existing models. For example, for companies trying to predict the interval between jet engine maintenance, the model may undergo major changes, and the use of jet engines has declined in recent months. The same is true for retailers, who have seen sales plummet in recent months.


Kurtz told CIO.com: “Some data changes very quickly, and historical conditions are no longer good at predicting indicators. This is quite common. Companies will have to reconsider algorithms because they never thought variables would happen before. Variety.”

Faced with such challenges, CIOs who are implementing artificial intelligence and machine learning discussed their work.


Health insurance companies use artificial intelligence to improve business results

According to Rajeev Ronanki, chief digital officer of American health insurance giant Anthem, the health insurance company has implemented artificial intelligence and machine learning solutions to handle many tasks ranging from predicting patient health trends to resolving service disputes.

By analyzing the health care data generated by patients with chronic diseases (such as diabetes or heart disease) over the years and comparing them with patients with similar diseases (or “digital twins”), Anson can predict the possible outcomes of treatment.

Artificial intelligence also helps Anson pay close attention to the progress of claims and other services, and discover potential customer problems in benefit claims rulings and other services. If Anson finds a significant difference, the customer service team will proactively contact the medical service institution or patient to explain the reason. Ronanki said that this kind of proactive outreach is essential to prevent tension in conflict relations. To this end, Anson analyzed historical data collected from millions of calls in which customers expressed dissatisfaction with the company’s services. Artificial intelligence generates scores that indicate the possibility of customers escalating complaints.


Ronanki said: “We will proactively contact customers and explain our decision, and try to provide specific background.”

In 2018, Anson hired Udi Manber, the former head of Google search, as its chief artificial intelligence officer, which shows that artificial intelligence has become very important in this company. Romanki said that under Manber’s leadership, each of Anson’s business lines has embedded artificial intelligence capabilities and skill sets, and cross-functional teams have developed applications. The goal is obviously to simplify the healthcare experience and make this experience more ” Personalization, efficiency and initiative”.

Transportation company uses machine learning to support package processing.


James Fairweather, chief innovation officer of Pitney Bowes, told CIO.com that Pitney Bowes is a century-old office transportation and mailing service provider and has been using artificial intelligence and machine learning tools extensively in the past eight years. The company currently uses machine learning software to predict when mail and parcel stations, including Android tablets and integrated printers, may fail. If the machine learning software in direct contact with the networked mail and parcel station detects a potential failure, it will arrange for an on-site service technician to inspect the machine.

Fairweather said that fixing the problem before the machine fails is critical to reducing the downtime of package transportation. And because machine learning software has gradually been able to accurately predict problems, Pitney Bowes can easily arrange maintenance services for the field service management system. “It provides customers with a great experience,” Fairweather said.


Due to the day-to-day shipping becoming more and more common, the consumer experience in transportation has become critical. Pitney Bowes also uses machine learning algorithms to optimize the return volume. For this reason, it is necessary to monitor the package route to identify continuous abnormalities during processing . Fairweather said that, for example, if a package is usually scanned every 4 hours in transit, but the second scanning window is missed, the algorithm will flag it.

Fairweather explained: “We built a data science model based on the normality of these activities to predict abnormalities during processing.”

Cranberry producer uses machine learning to strengthen operations

Before Ocean Spray embarks on the journey of artificial intelligence and machine learning, the producer of cranberry, grapefruit and other juices had to clean up the collected data for many years. The company’s chief digital and technology officer, Jamie Head, told CIO.com that the company has implemented a master data management strategy to improve the consistency and accuracy of information assets generated by its business units and customers.


Head said that Ocean Spray is using machine learning to comb through historical data over the past three years to assess sales growth trends and analyze competitors’ promotion methods to fill up any seasonal gaps that may exist. Head’s team is working with machine learning startup Visual Fabric to help understand how it can better gain insights from tracking expenditures to “drive the business.” The IT team shares these insights with the sales team to help them improve their marketing methods.

Ocean Spray also analyzes color, size, and other variables (including the soil and climate conditions of agricultural partners in Canada, Massachusetts, New Jersey, Wisconsin, Chile and other regions) to explore how to use machine learning to improve cranberries Quality of products.

Machine builders use virtual assistants to manage sales


Honeywell’s sales staff use artificial intelligence software to help prioritize meetings and manage sales leads to help the company’s avionics systems, engineering vehicles and other industrial machines win customers.

Patrick Hogan, vice president of business excellence at the industrial manufacturer, said that the software is actually a virtual assistant developed by Tact.ai, which can obtain information from Honeywell’s Microsoft Office 365 and Salesforce systems. Employees can use their smartphones to call or send text messages with Tact.ai assistants to check whether they are achieving sales targets as planned and to view metrics on how customers interact with business recommendations.

When the salesperson ends the meeting, the virtual assistant will ask them what the next step they plan to take. The assistant will also “push” notifications to users, indicating opportunities that may become obsolete. Hogan said: “Virtual assistants can help you master the situation in your department.” He added that the more you use this tool, the better you understand the workflow and preferences of each salesperson.

The virtual assistant has had a positive final impact on Honeywell’s sales channels, including more face-to-face meetings, which has improved the sales, sales conversion rate and profitability of each seller. He is actively urging the company’s 9,500 employees More people in use the tool.

Artificial intelligence promotes personalization of business services

According to Todd Hale, CIO of Office Depot, Home Depot is investing in machine learning capabilities to gain insights into customer preferences and recommend products more accurately.

The company, with annual revenue of 11 billion U.S. dollars, strives to expand its business service department (including its CompuCom technical service department), while reducing its dependence on office supplies sales, so it has carried out data analysis. B2B sales contribute more than 60% of Home Depot’s revenue. The company uses advanced artificial intelligence/machine learning technologies (such as XGBoost and Random Forest) to segment customers into multiple roles and predict customer churn rate, customer lifetime value, and product affinity.

Hale said: “In the field of e-commerce, we use the deep learning capabilities of Apache Spark and Analytics Zoo on BigDL to provide user-based real-time product recommendations and develop cross-selling and up-selling models.” He added, ideally , Which will help Home Depot develop “tailor-made products and services.”

Leave a Reply

Your email address will not be published. Required fields are marked *