Blog

Algorithms for Personalized Recommendations

Users do not always leave behind enough personalized information along their customer journey. For instance, new customers can be acquired or existing customers might browse an e-commerce website without being logged in. Non-personalized recommendation systems, such as those based on proposals for products frequently purchased together, still offer recommendation opportunities for companies in this case. However, the more individually these are tailored to the customer, the better.

Anyone who evaluates data using well-known visual analytics tools such as Tableau or Qlik Sense wants to create, quickly reach the limits of applications — at the latest as soon as he works with unstructured or incomplete data sets. To avoid frustration, we tested both tools for their data preparation options and created a short list of best practices.

How much does it cost me to use the tool? Who should work with it: marketeers or IT professionals? And can I also manage my data in the cloud? Many factors play a role in choosing the right tool for visual analytics. b.telligent consultant Daniel Erlhöfer explains the technical side of Tableau and Qlik Sense in this blog post: What are the differences in the data preparation of both tools?

Starting Situation

One may expect widespread approval when describing the objective of modern campaign management today as follows: The focus of marketing must be the individualized and consistent interaction with the customer via his/her preferred channels - self-evidently containing relevant messages at the respective best time. While this statement is not very surprising nowadays, it still shows that behind each of these adjectives, there is a potential source of errors for a customer-oriented campaign strategy.

Neural Networks for Tabular Data: Ensemble Learning Without Trees

Neural networks are applied to just about any kind of data (images, audio, text, video, graphs, ...). Only with tabular data, tree-based ensembles like random forests and gradient boosted trees are still much more popular. If you want to replace these successful classics with neural networks, ensemble learning may still be a key idea. This blog post tells you why. It is complemented by a notebook in which you can follow the practical details.

More than a year after introduction of the general data protection regulation (GDPR), many enterprises still find it hard to reconcile the topics of data warehouse (DWH) and data privacy. Customer-centric data modelling prevailing at enterprises poses a special challenge here. It leads to major conflicts with many GDPR requirements in virtually any data-driven process. But why is it so hard to unify the two topics?

As a business intelligence package, SAP BW provides many opportunities for efficient reporting – but also contains numerous barriers which significantly slow down performance. Using the example of report filters available in the application, this article shows how the smallest adaptations impair SAP BW performance, and how efficient settings for filtering options can improve results.

Reinforcement learning has a bad reputation for being extremely data-hungry – so data-hungry it can only realistically be trained in simulation-generated data, e.g. in a computer game. We discuss how this can be cured using Bayesian Statistics, using an easily accessible small example. In the second part of this blog series, we see how this can be done in practice using TensorFlow Probability, a hot new tool from Google.

In previous years, there were repeatedly reports and rumours about the future of SAP BusinessObjects.

Some interesting new details have emerged in the meantime:

News on SAP Business Objects BI 4.2 SP07

On 04.03.2019, SAP issued support package 07 for the SAP Business Objects Business Intelligence Platform 4.2 (SBOP BI 4.2). According to SAP, this is the last feature pack for release 4.2 . Though support package 08 is planned, it will constitute purely a maintenance release containing "only" bug fixes but no new functional extensions.

At last: The starting shot for digitization strategy has been fired, and marketing is meant to be an important component of this strategy. In the course of implementation, a decision was made to use a customer data platform (CDP) to form a 360-degree customer view and automate marketing journeys in a real-time environment. This decision, in addition to an investment which is not small, has significantly influenced marketing processes and areas (online, e-mail, e-commerce and CRM) at an enterprise. In this third and last part of my CDP-series I would like to focus on the appropriate selection and evaluation of the suitable CDP-tool.

Unlike marketing automation solutions, a distinction is made not according to functional scope such as enterprise or best-of-breed, but according to the customer data platform's functional focus. In other words, the right kind of CDP can be selected already in advance, depending on how requirements are defined.

A total of three categories of customer data platform are distinguishable:

• Data CDP
• Analytics
• Engagement CDP

This end-to-end and step-by-step quick demo shows you how to connect an existing on-premises data source to a modern cloud data warehouse solution such as Snowflake:

• Easily
• Fast
• Securely, and
• Without having to write a single line of code.

Before HANA, exception aggregations in SAP BW more frequently posed a challenge at runtime, this also applying to the distinct count operation. This operation is used, for example, to obtain a customer count from orders.

Earlier, distinct count operations were often implemented as follows:  A calculated key figure with a value of 1 was established, and additions then performed via exception aggregation. However, what worked well in environments without HANA currently causes the pushdown to stop functioning. This can prevent calculations from being optimally performed, depending on settings.

A solution is therefore required for HANA environments: Accordingly, let us concentrate next on optimal implementation of distinct count with business warehouse on HANA or BW/4HANA.