Positioning:
SAP Predictive Analytics 2.0 will be of particular interest to existing SAP users, and combines SAP Predictive Analysis software and SAP InfiniteInsight. This makes it very suitable for customer focused analytics, and SAP clearly understands that creating the actual model is just a small part of the journey. To this end they provide extensive facilities for data preparation, visualization, data modeling, scoring and most importantly model management. With its application suites used in many large corporations around the world, the analytics capability will be welcomed by those wishing to build predictive models for anticipating customer behaviour, and with the emerging Internet of Things (IoT), many businesses rich in hard assets will want a platform for analytics on data coming from sensors and devices. With its HANA database technology SAP can offer a complete solution to the real-time analysis of events originating from devices, customers or any other activity.
Business Application
The obvious benefit for SAP users is the complete integration of the predictive analytics process with their SAP infrastructure. A variety of businesses us SAP Predictive analytics including:
- A very large insurer with 31 million customers uses SAP Predictive Analytics to analyse customer behaviour and conduct more targeted customer marketing. This means they have been able to generate 80% of sales from just contacting 50% of the customer population.
- Product return rates were a major concern for a home shopping business. By using a combination of SAP HANA and Predictive Analytics they were able to significantly reduce returns, and more accurately target customers with offers.
- Customer churn rates were a concern for a large mobile phone operator, and by using SAP Predictive Analytics churn rates have reduced and customers are presented with more relevant up-sell and cross-sell opportunities.
There are many more examples of SAP customers deriving value from their predictive analytics technologies, and the application range across finance, planning, customer service, predicting product failure, machine maintenance and many others.
Technology
There are essentially three predictive model avenues that can be pursued within SAP. The InfiniteInsight platform is very much customer focused and employs structural risk minimization techniques to ensure models are reliable. The open source language R can also be used to create models addressing virtually any business problem, and HANA comes with its own high performance algorithms.
The data preparation task, which is both time consuming and tedious, has been addressed to a large extent in the new release. Most tasks can be accomplished without recourse to code. Data visualizations are important in the early stages of model development for gaining a feel for data. The new visualization capability can access data from multiple sources and provides a rich set of visualization tools. The data modeler supports both business users and data scientists build predictive models, and includes many automated features. The model management capability is worthy of special mention, allowing automatic scheduling of model refresh and support for what-if type questions.
Network and link analysis is also supported, and is typically used to explore links between customer and participants in social networks.