Well, if I said the truth, it would make me sound really old, so I’ll say 30+ [but it’s actually over 40…]. I started off working in the operations research department of Unilever. We built what was then the world’s largest linear programming optimization models for animal feeds production planning. I was also a senior lecturer in statistics and operations research at the University of North London.
What’s your role today?
I’m SAP’s VP and Head of the Center for Predictive Analysis. I act as a central source of information across the different SAP predictive initiatives…
Could you give us an quick overview of the different predictive initiatives at SAP?
The main four are:
In-database predictive analysis library in SAP HANA -- it’s a collection of over 25 predictive algorithms that can be run directly in-memory.
R integration for SAP HANA -- R is a widely-used open source programming language for statistical computing and graphics. The integration enables the SAP HANA database to process R code in-line as part of the overall query execution plan.
SAP Predictive analysis, a user tool for the definition, visualization and running of predictive processes, either on HANA or non-HANA sources.
Embedding of predictive functionality within SAP’s industry and line of business applications – such as customer engagement intelligence, where it’s used for segmentation analysis, customer lifetime value analysis, etc.
In addition, there are number of more recent activities in the predictive space:
SAP’s acquisition of KXEN. KXEN makes predictive easier for business users, by taking a class of problems, such as segmentation, association, and time-series analysis, and providing a generic algorithm for each type of application. The key advantage is that business users don’t need know which algorithm to use when.
Announcement of our partnership with SAS. We have many joint customers that have rich experience with SAS, so it makes sense to bring the benefits of the HANA platform to the SAS applications. We’re initially focusing on five business applications, and two SAS algorithms, for time-series analysis and social network analysis. The type of integration is analogous to PAL – they have been embedded in HANA, so the data remains directly in memory.
What’s the strategy going forward?
We want to bring the benefits to predictive analysis to every aspect of our customer experience – providing deep and broad predictive functionality, not just for data scientists, but also for business users.
What about “Big Data” and predictive?
Obviously, with the explosive growth in data collection -- for example machine data – we have more information about what has happened and is happening, which makes it easier to predict what might happen. With more granular data and new data sources, we can build predictive applications that previously were inconceivable. It also enables new approaches. For example, in the past, we had to use sampling to reduce the size of the data set in order to do the predictive analysis. Now we can use the whole data set, or use more modern approaches such as ensemble models, where we produce many different models in order to explore the “total model solution space.”
If I’m already an expert in predictive, where do I go to get more info?