How do I implement machine learning

Three fields of application for machine learning in companies

In the meantime, powerful machine learning algorithms that can be easily implemented in companies today have proven themselves in three ways. An overview of the possible uses.

As an engine that recognizes patterns, machine learning enables the further automation of existing business processes.

Process automation: this is how machine learning works

  • Large amounts of high-quality data are required for machine learning. In most organizations, this data can be found in existing business applications for finance, logistics, and sales. This data has already been recorded, cleaned and stored for a long period of time. So there is a lot of data that can be used to create meaningful, useful forecasting models.
  • Machine learning works best when there is a well-defined decision to be made, as it happens several thousand times a day, that involves a small number of variables, and where errors are clear and can be quickly corrected to further the algorithm to improve. An example: “Which of these transfers corresponds to this invoice?”. Machine learning is much better for such a question than "How can we improve lung cancer survival?"
  • Machine learning is easiest to implement when the decision can be seamlessly automated as part of an existing business process without the creation of new processes or cultural changes.

Some examples of automatable processes:

  • Extraction of relevant payment or order data from unstructured invoices, forms and emails (e.g. product name, amount, currency, payee, address, etc.)
  • Classification of transactions for compliance with tax guidelines
  • Predict when to renew usage-based contracts
  • Forecasts and actions in the event of stock-in-transit delays
  • Calculation of the optimal time between inventories to ensure synchronicity with automatic systems
  • Forwarding customer service requests to the most appropriate team

These seemingly boring machine learning use cases are by far the greatest real-world opportunities for business benefit today. McKinsey has calculated that around 43 percent of financial processes can be automated with the help of AI. Gartner says AI saves half a billion people two hours a day of action and decision-making time this year alone. So there is huge potential when you combine machine learning with sensors, IoT and other technologies.

More intuitive user interfaces

Recent advances in machine learning have greatly contributed to the ability of computers to decipher and understand human language, writing, and commands.

New service chatbots, for example, make it easier for customers to find information and carry out simple transactions via voice or chat interfaces. Machine learning algorithms scan a wide variety of products and technical documentation and automatically answer common questions. The first implementations show: The use of chatbots to answer basic questions accelerates customer conversation, increases customer satisfaction and significantly lowers costs.

In companies, digital assistants can help with everyday processes.

Within companies, digital assistants can help with everyday processes, especially in connection with central business processes such as procurement, human resources and budgeting. Instead of clicking through a complex interface, employees could tell the digital assistant that they would like to take a vacation next week or ask them how the current actual costs in your department are compared to budget.

In a work environment, digital assistants have access to large amounts of contextual information that can facilitate or even anticipate such an exchange. The system knows how many vacation days you are still entitled to and what budget you are assigned to. With the help of machine learning, the system can even identify exceptional circumstances and draw attention to them without having to search for the relevant information: “Based on your current reservations and forecast trips, you will exceed your travel budget by 30 percent this quarter - you want to review your trips or make the finance department aware of it? "

Disclose and optimize processes in a way that was previously not possible

Machine learning can help efficiently process data that was previously too complex or expensive to analyze. This gives you insights into processes that can be optimized in new ways.

Some examples:

  • Predicted maintenance:With detailed sensor data and algorithms, the first real signs of problems in parts or in machines can be identified. This saves enormous sums of money, because otherwise the parts would be replaced regularly, regardless of whether they are worn or not, or - what is even worse - there is a wait until the parts are defective and production comes to a standstill. These technologies are even being used by humans, such as professional athletes. By monitoring players' vital signs during training, it can be ensured that they are performing at their best and the risk of injury that would result in player failure is minimized.
  • Image analysis and monitoring:The performance of the new deep learning algorithms, which are used to interpret and understand complex images, has increased particularly sharply. This has opened up a multitude of business opportunities. For example, an oil company can scan their barrels to make sure they are correctly and clearly labeled. Sponsors of sporting events can receive detailed analyzes of how often, how long and where their logo appears when a sporting event is broadcast. This helps them to optimize the reach and to recognize whether their investments are bringing the desired success. Companies that have complex catalogs with many different products and variations can use the algorithms to quickly identify the right product in a photo, regardless of whether the company is selling office supplies, tires, or jewelry.
  • Text analysis and classification:Machine learning can be used to extract text and images from electronic documents. This information can then be classified so that it can be analyzed more easily than ever before. Use cases include texts on potentially fraudulent claims for insurance companies, sentiment analyzes for customer loyalty, classification of drug interactions based on research documents, and much more.

Find out how companies can benefit from machine learning.

More from solutions

SAP launches free training courses for students

Newsbyte - SAP today announced the SAP Learning Hub, Edition for SAP Next-Gen, a free training offering focused on students and faculty. From May 2021 ...

Continue reading

Kaeser compressors: using smart engineering to create a precisely fitting compressed air system

Feature - The development of future-oriented industrial plants poses major challenges for machine builders. In order to accelerate the design process of tailor-made compressed air systems, Kaeser Kompressoren relies on a collaborative smart engineering platform ...

Continue reading

Hannover Messe: Automated material supply in intralogistics through Industrial IoT

Feature - reducing costs, increasing productivity - that is currently the focus for SAP customers. At the virtual Hannover Messe 2021 it will be shown how SAP Industry 4.Now customers ...

Continue reading