Embrace Machine Learning with SAP in 2019—With the Right Set of Data

In 2018, SAP used both carrot and stick to promote the move to the cloud by tying their most advanced offerings to SAP Cloud Platform—Analytics and Machine Learning, for example. In 2019, SAP will continue to emphasize that the “Intelligent Enterprise”—SAP’s vision of integrating applications and enhancing them with new technologies—is a necessary stage in a company’s development. Companies that ignore it, they imply, risk irrelevance or, perhaps worse, featuring in a Business Insider listicle of failed businesses. 

At SAP’s fall technology conference, SAP TechEd in Las Vegas, it seemed that over half the session offerings covered Machine Learning, Artificial Intelligence, and Analytics. The topic of Machine Learning/AI is all the rage across many industries and novel applications of the technology continue to bestow enviable competitive advantages on companies with enough creativity and talent to leverage its power. 

SAP’s intent is to bottle that power. SAP has convinced me that they’re invested in workable Machine Learning in SAP HANA and on SAP Cloud Platform. They provide their own models, but I’m most excited about the ability to drop in my own neural net model that I’ve built in mature, proven technology like TensorFlow or PyTorch. Better yet, I can use one of the many free, high-quality models available online from open source consortia or companies like Intel. Intel offers its Computer Vision framework for free, and it provides dozens of advanced deep learning models that will run in real-time on a CPU. That’s a great place for companies to start experimenting with Machine Learning and AI.  

Another compelling reason to experiment before developing new products and processes with Machine Learning: you can avoid applying it to problems that don’t require it. For example, I find the SAP Analytics Cloud tool compelling, with its capabilities in model building, insight and discovery, and visualization all in one place. That doesn’t mean the tool’s AI engine needs to be applied to every task that is just fine—and even better served—by traditional regression and forecasting methods. 


Privacy and Consent in the Machine Learning Era 

One subject that needs more amplification for 2019: privacy and consent in the era of Machine Learning. Customers are valuing trust in consumer industries more than ever—and there is legislation such as GDPR to back that up.  

It is challenging to build a Machine Learning model—and the more data one collects, the better the chances of the model producing something usable. But the even bigger challenge is to identify where the value of Machine Learning is eclipsed by loss of goodwill due to invasive data collection and use practices. 

SAP has demonstrated a Computer Vision application a type of Machine Learning) that interprets a customer’s facial expressions for a sales rep during a video call. How that benefits the salesperson is one question, but the other that we must examine is how a customer would be feel about that happening. As we build technologies, we must keep that end-consumer in mind. If we lose their trust, it will be hard to get it back.  


Building a Data Foundation 

The most sophisticated neural net in the world can’t be effective without good data. That’s why before embarking on any advanced analytics, machine learning, or artificial intelligence journey, retailers and those in the consumer industries must be sure that they have a good data backbone and are collecting the right data for analysis. 

As we advance through 2019, remember that data is still the key. At /N SPRO, we help customers implement solutions such as SAP Customer Analytics Repository which can serve as that initial data foundation from which to build advanced systems incorporating machine learning and building towards SAP’s Intelligent Enterprise.  


Contact us to help build your data foundation.  

By | 2019-02-01T16:28:00-05:00 February 1st, 2019|Categories: Blog|0 Comments

About the Author:

Leave A Comment