What’s New for Predictive in SAP HANA 2?
HANA 2 is out! If you want to delve deeper and see working examples of what’s new for the predictive analysis library (PAL) with SAP HANA 2.0 SPS00 then you’ve found the right place.
This release brings a bumper crop of new predictive algorithms to the PAL as well as significant enhancements to existing capabilities including optimized parallel processing for scoring partitioned data.
There are quite a number of new algorithms encompassing association analysis, statistics, time series analysis, classification and regression.
One particular highlight is sequential pattern mining for association analysis. The PAL has supported association analysis for a number of years via algorithms such as apriori and FP-growth. Sequential pattern mining extends these by supporting the concept of time when building association rules. So, for example, customer purchases can now be analyzed over a given period time which can help identify more useful cross selling recommendations.
Fast Fourier Transform is a well established way of analyzing signal data over time in order to identify it’s constituent components. It’s a bit like analyzing a cake in order to find the recipe. Fast Fourier Transform can add significant value to real-time IoT scenarios!
You probably already know that one of SAP HANA’s most powerful capabilities is the ability to partition data on a single-node or across multiple-nodes of a SAP HANA landscape in order to provide scalability?
With HANA 2, it’s now possible to run parallel execution of selected PAL functions against a partitioned table simply by adding the following clause in SQL Script:
WITH HINT (PARALLEL_BY_PARAMETER_PARTITIONS ())
The main use-case is to run scoring against a trained model from one of the PAL supervised learning algorithms such as decision trees or random forest. Given a partitioned data table, the parallel execution of the scoring function is initiated on each data partition, sharing the same trained model. This is fully supported for both single-node and multiple-node SAP HANA landscapes and really allows you to exploit the full potential of SAP HANA.
To help get started with all these new capabilities the SAP HANA Academy has created hands-on video tutorials. In case you’re new to SAP HANA and the predictive analysis library, the getting started tutorial has been extended to include an full overview of predictive with SAP HANA – so there’s no need for prior knowledge of SAP HANA 1.0 in order to get going.
Here are direct links to the hands-on video tutorials published so far:
Intro and Overview
Sequential Pattern Mining
Gradient Boosting Decision Tree – Model
Gradient Boosting Decision Tree – Predict
Linear Discriminant Analysis – Fit
Linear Discriminant Analysis – Classify
Linear Discriminant Analysis – Project
Finally, if you’re interested to learn about what’s new with HANA 2 in general check out the following blog.