 |
 |
|
|
Home
Our scoring system Score™ 7.0 is a combination of neural networks and traditional techniques of statistical preprocessing and analysis of customer data. Thereby, the whole transparency of statistical analyses is achieved together with the full non-linear strength of the analysis of neural networks.
All analyses are highly transparent, i.e. the influence of each variable on the total result can be visualised and analysed - a feature which makes Score™ 7.0 not only "boardproof" but also effectively immune against possible faults in the source data. The full transparency of the results in the use of non-linear neural networks is unique in the world.
Compared with other scoring systems, Score™ 7.0 accepts a huge number of variables per customer record. A pre-selection of variables is therefore not necessary. Instead, Score™ 7.0 automatically exploits all of the informational content of detailed customer data, without requiring any presuppositions. A use of up to 10.000 variables per customer record is the current standard.
The use of neural network techniques makes Score™ 7.0 very unsensitive for defective or missing customer data. Besides, exceptions (i.e. customers with an extremely deviant or untypical behaviour) are identified automatically and dealt with separately. The sampling of huge data inventories via ODBC or ASCII ports is fully integrated in Score™ 7.0. This makes any use of additional preprocessing tools redundant.
If necessary, a very high processing speed is possible. It is no problem to achieve an analysing time of only a few hours with Score™ 7.0. The scoring speeds will then be at a typical several million data records per hour (with 5.000 variables per customer record, on a common PC). With Score™ 7.0 you will have an unproblematic and extremely speedy application even when it comes to huge numbers of customers. Thus a very current application of analyses is possible - which is of crucial importance for a success in direct marketing.
Score™ 7.0 runs on every common PC under Microsoft® Windows® and 1 GB RAM. This means that Dr. Imhoff Business Consulting can carry out the complete data processing locally, i.e. in the customer's office. The software is able to use the full capacity of computer systems automatically, either of multi-processor computers or of computer networks. This is made possible by strict multithread programming and parallelisation of all processing intensive processes.
Score™ 7.0 has a fully integrated ETL tool (ETL = Extract Transform Load), which reduces the amount of data
modeling for the data mining process to only a few days work. ScoreETL™ creates a virtual data warehouse that is
optimized for the data mining process. Any data base programming or building of a data warehouse or data mart is
now superfluous. Data mining projects become considerably faster and cheaper this way.
In any data mining project the collection and preprocessing of data takes at least 80% to 95% of the complete project time. To dissolve this bottle neck and completely relieve the IT divisions of our customers, we developed ScoreETL™ and integrated it into our data mining system. By directly accessing the operative systems of our customers we reduce the effort on side of our customers to nearly zero thereby accelerating data mining projects vastly.
The complete data mining system Score™ 7.0 is implemented in an object-oriented way from the start, in other words, there is no add-on object orientation as is the case with many other software packages. The central routines (about 95% of the code) are also written independently of a platform, so that a portage from Windows to, e.g., Unix derivates (Solaris, Linux, AIX etc.) is easily to be realised, because only the user interface needs to be customised. The object orientation and independence of platform allow an easy maintenance and extension of the software in the future, as well as a quick customisation to future electronic processing surroundings.
Besides the wellknown ODBC-access to all familiar database systems, the software can be used as a standalone data mining server. Being a standard HTTP server the system can provide its complete data mining functionality to any other software system in the intranet or in the internet. The data interface of Score™ 7.0 is freely programmable as a plug-in and can even be exchanged during runtime without stopping the data mining server. As a standard data interface Dr. Imhoff Business Consulting provides a freely configurable XML-interface, which allows direct access to all modern CRM systems. Being a plug-in the data interface can be adapted to any proprietary data exchange format if necessary.
Score™ 7.0 is optimized for the use under Microsoft® Windows®. If used on a simple PC or on a server cluster, Score™ 7.0 can be easily adapted to any requirements. The extensive scalability of Score™ 7.0 allows the implementation of very small systems with very low hardware and software expenses. Upscaling to very high system power can be achieved without any adaptation of the software itself.
Using multithread architecture Score™ 7.0 automatically supports the full computing power of multiprocessor
systems. At the same time the server structure of Score™ 7.0 allows the full exploitation of the power of
server clusters per load balancing. For a further increase in computing power Score™ 7.0 internally optimizes
all accesses to one or several database systems. The software achieves a minimization of the number of database
transactions, and an automatic load balancing if more than one database system is used simultaneously.
With Score™ 7.0 we can implement data mining systems of any size: it can be deployed on a simple PC, on a multiprocessor system or a high-capacity server cluster in exactly the same way.
Score™ 7.0 has been tested in several comparative analyses by comparing it to concurring companies' products. Score™ 7.0 has scored far ahead of any other product, with the actually achieved sales figures far better in all comparisons than those of the concurring companies. Score™ 7.0 is currently being used as a tool for the selection of customers by several big international mail order companies and non-profit organisations.