Predictive & machine learning has been around for decades. However, new market forces are changing the landscape and creating new opportunities for its powerful application. Organizations like yours can identify untapped opportunities and expose hidden risks buried inside vast amounts of data—all in real time.
In a broader predictive & machine learning context, there are three categories of personas
- Data scientists
- Business analysts
- Business users or Decision-makers
In many cases, both the data scientist and the business analyst are the producers of predictive & machine learning models, whereas business users are consumers of the predictive & machine learning models directly or indirectly. A great example is a call center agent when making a next-best offer to a client is unaware that machine learning is feeding the script they’re following. Automation of predictive & machine learning process critical for those personas.
Predictive & Machine Learning tools and libraries consist of standard algorithms that primarily fall into six main categories
- Classification – Who will churn, commit fraud, or buy next week/next month?
- Regression – How many products will a customer buy next month/next quarter?
- Segmentation or Clustering – What are the groups of customers with similar behavior or profile?
- Forecasting – How much will be the monthly revenue or number of churners next year?
- Recommendations – What is the best offer or recommended action for a customer or internet user?
- Association or Link Analysis – How are the customers and products related to each other?
To solve these, you could approach
- With one or more algorithms and build models with a choice of your tools and algorithms
- Automate the entire process.
Most previous attempts to scale machine learning to massive datasets failed miserably due to the nature of the first-generation multi-algorithm workbenches and most previous efforts on managing an enormous number of models in operations failed because it demands high degrees of individual skills using first generation tools.
With Automated Predictive Library (APL), you can automate of a massive number of models build and applied on massive datasets and leverages existing skills of the current enterprise.
APL is designed to solve time series, clustering, binary classification, regression and recommendation use cases. It is based on the automated machine learning engine (KXEN). APL is designed generically to support variety line of business and industry-specific needs. One of the critical features of APL is wide-dataset support – data sets that contain up to and over 15000 columns (truly Big Data). You can quickly develop and deploy predictive & machine learning models with HANA based applications.
SAP HANA delivers the broadest set of advanced analytical processing engines to understand and analyze all data types. SAP HANA advanced analytical processing engines provides predictive & machine learning, spatial, text & search, graph, streaming, time series, and document store capabilities.
With SPS 04 (released on April 5th, 2019), Automated Predictive Library (APL) is now part of SAP HANA. What it means:
- Existing SAP HANA Predictive Option or Enterprise Edition customers have entitlement rights to use APL
- Future customers of SAP HANA Predictive Option or Enterprise Edition customers will include APL
This is a game changer because our customers and partners can easily build and deploy intelligent applications on one copy of the data – without duplicating data to achieve acceptable performance, moving data around to create unified views, or using multiple programming models to process different data types.