Augmenting Business Intelligence with Machine Learning

Augmented Business Intelligence brings new capability to an old set of business problems—delivering faster, more complete answers automatically and putting data science techniques in the hands of business users.

SAP Analytics Cloud puts extreme power into business users’ hands by providing insights never before available in business intelligence (BI) products. With a large investment in augmented capabilities, we capitalize on the vast quantity of data across the enterprise—delivering deep analytical capability. Combine this with the flexibility and simplicity of SAP Data Warehouse Cloud, and we are opening up a world of new possibilities for BI.

An Established Space with New Potential

Business Intelligence is a mature industry. BI, at its core, is about understanding what’s going on in your business. Organizations use BI to monitor for problems, get at root causes, and push data-driven decision-making in the enterprise.

With the emergence of self-service in the last decade, BI has focused on tools to allow users to search for answers themselves and present them to others.  Augmented Business Intelligence dramatically evolves the self-service concept by automatically mining for patterns and providing answers in a  presentable way. With this, users can focus on really understanding and actioning the results rather than tedious and biased data exploration or presentation-building.

Complex Insights for Everyone

Business intelligence products are designed for use by business analysts while predictive analytics tools are made for data scientists. In a conventional environment, business analysts using BI solutions may have access to the output of predictive analytics, but in a purely downstream way. They can’t control or modify the analysis.  They understand the business but not the statistical methods or machine learning technology.  The data scientists who do the analytics are at a premium because their skills are highly specialized, and their numbers are limited. This leaves insights unfound and patterns unseen by the decision makers of the business.

For example, if you want to focus on improving a particular business metric such as employee attrition, you could classically kick off a data science project.  That data science project is likely to take multiple months and also requires premium resources and skills. First, requirements would be gathered via interviews between the business and the data science team; then data would be prepared, models would be created and trained, and the process would continue with incremental improvements in all stages until the very end—where a report would be delivered to the decision-maker.  Valuable time is lost. No action was taken on your employee problem, and no other projects were able to be analyzed.

Instead, why not allow an automated system to tell you what the key drivers are just from looking at historical data? The business user can help guide the system. For example, they understand that even if attrition is being caused by local housing prices they can’t affect it, so they should look at other things.  They can also further iterate by asking intelligent follow-up questions on other related factors like training spend and manager trust.

Not only would you get your answer quickly and easily, but you then free up your data scientists to work on higher-value problems.

The Trust in Automating Data Science

By now, you have recognized that our augmented BI approach automates data science techniques. But in order to put this into practice and really benefit from it, let’s explore how we do that, because trusting the results is as important as finding them.

In the example where we want to identify what is behind employee attrition, our system will use industry best practices to identify the best predictive technique to apply—in this case by building a regression model.  We would require the historical employee data and we would automatically build and compare a multitude of regression models to find one that has good quality, robustness, and accuracy.  Behind the scenes we tune parameters and validate it against common risks such as overfitting.  When a model can be found with sufficient quality, we will present the resulting insights back to the user with simplified metrics.

Because we disclose the key facts about the users’ data and allow them to explore the relationships found by the model they typically establish trust by simply validating what they already understand to be true in their business. This adds credibility to the new insights found.  We also provide the more complex metrics behind the model to gain trust by their partners in the data science community.  Ultimately, by automating the process and providing answers in minutes rather than months we can drastically affect the likelihood of an employee leaving.  This is only possible because automated modeling happens right inside the tool and offers this to a broader set of users.

Augmented Business Intelligence and SAP Data Warehouse Cloud

At the foundation of good decision making is trust in the data itself.  This is why SAP Data Warehouse Cloud serves as a single source of truth for the business. This trusted information is critical infrastructure for augmented business intelligence which avoids a common pitfall of garbage-in-garbage-out results. SAP Data Warehouse Cloud supports augmented BI by ensuring that ultimately the data is credible and good quality.

A critical feature of SAP Data Warehouse Cloud is the semantic layer that maps business concepts to the underlying data architecture. Effectively, it translates data concepts into business language. This is a critical improvement for analytics generally, but it is especially useful in the context of augmented BI.   As we automate more and more of the mundane aspects and focus on delivering powerful information to users, they need to understand the results; the relationships in information must be well modeled to allow us to extract truly credible answers.  With augmented business intelligence working in conjunction with SAP Data Warehouse Cloud, you have an environment that understands your business, understands your questions, and is ready to answer them before you ask.

A Modern Approach to BI

More than 30 years since we defined the term Business Intelligence much has changed and much has stayed the same.  It is still important that we tell users what is happening in their business and why it’s happening. But augmented business intelligence addresses those challenges in a completely new way, leveraging advances in machine learning to help your business get all the information.

With the big picture in your hands you truly can make confident decisions.

Learn More

Learn more about intelligent data warehousing on the SAP Data Warehouse Cloud product page.

VN:F [1.9.22_1171]
Average User Rating
Rating: 0.0/5 (0 votes cast)

35 Views