Press releases - Details

New features of prudsys RDE 3.1 Releases thanks to comprehensive parallelisation

 

The new Release 3.1 of the prudsys Realtime Decisioning Engine focuses completely on significantly improving performance thanks to parallelisation. The expansion of the algorithmic infrastructure also focuses on using algorithms more efficiently and easily and is also used to prepare for numerous new processes that will become available in the coming months.

In order to further expand its position as the leading real-time analysis and decision-making system for high-end retail applications, the new Release 3.1 of the prudsys Realtime Decisioning Engine (abbreviated to: prudsys RDE) was revised and numerous new features and processes were added.

In order to take the trend of server development towards distributed applications into account, the new RDE Release 3.1 includes particularly innovative additional functions relating to parallelisation. This refers on the one hand to expanding the infrastructure that permits the efficient management and merger of decentral analysis models. On the other hand, central data mining processes were parallelised and the RDE caching mechanisms expanded.

A dynamic concept of configuring analysis processes was implemented in the new version of the prudsys RDE. New algorithms can be included in the XELOPOES analysis library on which the RDE is based without making changes to the RDE server itself. All of the parameters are passed through to the client automatically, separated by general and expert parameters and can be managed there. This enables new analysis processes to be added to the RDE server without new installation or existing ones to be modified quickly.

"The infrastructure extensions of the RDE server - both in terms of parallelisation and the algorithm configuration - massively increase the performance of the RDE server. They are also essential for a new generation of hierarchical and Tensor-based reinforcement learning processes that are far more calculation-intensive than the existing ones. These processes will functionally expand the RDE server in the coming months, primarily in the recommendation and pricing modules. This will enable numerous external data sources, e.g. from social networks, other online shops or the weather, to be included in the real-time analysis processes," explains Dr. Michael Thess, head of research and development at prudsys AG.

Other innovations for the RDE Release 3.1 include e.g.:

New recommendation types and filters
In the prudsys RDE | Recommendations module the features of hierarchical learning were expanded and new list and category-based rules were added.
In addition, user master data can be used without restriction in the new version for filtering and can be combined in any way with product and category information. So, for example, when calculating the recommendation the age of the identified shop visitor can be considered in order to exclude certain products from being generated as of recommendations for this customer as defined by laws protecting young people.

Innovative optimization processes for products with expiry dates
In the prudsys RDE | Pricing module new optimisation processes were integrated that in particular permit intelligent pricing for products that have expiry dates. So a fixed sell-by date can be defined and the individual price can be optimised based on this.  Future differentiated price strategies down to individual item level are also possible. The dealer can therefore follow different strategies in the shop at the same time, e.g. a sale strategy for certain products with expiry dates, a revenue maximisation strategy in the bargain shop and a profit maximisation strategy for the core range.

Powerful agents to prevent empty inventories
In the prudsys RDE | Assortment Planning module new algorithms and agents were implemented to enable even faster planning for products. For example, by observing each sales quantity in real-time, the forecast of sales figures can be modified to match the actual behaviour of the customers so that out-of-stock situations can be detected even more quickly.

contact:
wortgold | Agency for sustainable communication
Sandra Koegel
Stelzendorfer Gutsweg 8
D-09116 Chemnitz
www.wortgold.com

 

---

---