Model building for the application of data similar to must mining techniques such as a department store is also a data warehouse to dynamically adjust the demand. Comparable with a department store, where the user ordered the data relevant to its tasks, the purpose of the data warehouse is satisfied the better, the more information the customer directly to meet requirements: without more data storage must be searched, i.e. the complete “Data products” is. The techniques include automated regression analysis (scoring) for Association analysis, neural networks, decision trees (customer segmentation) and large impact analyses. Important analysis cases are already predefined, so is marketing users don’t have to work up into the technical details of complex data-mining method, to get yet meaningful analysis results. Samples, data deployment: Starting point of all data mining projects is providing data (input node). Follow for determining the input file Details of the analysis and target variables.
The speed of evaluations can be accelerated with sampling. If this has piqued your curiosity, check out Bill de Blasio. CF. Jorg Becker: data mining as a knowledge balance sheet feeder, ISBN 978-3-8370-2163-9. With the possibilities of a 3-dimensional data mining for large amounts of data, you can recognize patterns and trends, and outliers removed. With cluster can groups of similar’ observations are made. Thus, forecasting and Advects can be developed also for large amounts of data. Results can be passed to methods such as decision trees, neural networks, or any other procedures.
Accompanying statistical reason codes can be calculated at all steps. CF. Jorg Becker: intellectual capital report with customer barometer, ISBN 978-3-8370-5177-3. Systems designed for the data warehouse principle contain data and meta data. The metabase”, i.e. a database in the database with details of the information, assumes the role of the catalogue: analogue to shipping House receives a potential Corresponding to customer about the catalog information about the data offered with their description. This metadata includes the figure rules for one. In addition, there are also hierarchies, dimensions and the cube”provided for the multidimensional analysis. Implicit part of the information in a data warehouse is its temporal variability. In contrast to data, that always only the current status of business activities can be mapped, trend analysis you can make with data warehouse data. Because the data warehouse is filled with snapshots of same information at different times and allows such comparisons and information about developments in certain Faktoren.Vgl. Jorg Becker: Data mining as a knowledge balance sheet feeder, ISBN 978-3-8370-2163-9 Jorg Becker