Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations
The KDD process for extracting useful knowledge from volumes of data
Communications of the ACM
| Hi-index | 0.00 |
GainSmarts is a data mining system in support of database marketing decisions, encompassing the entire range of the KDD process, including data import, exploratory data analysis, transformation of variables, feature selection, data mining, knowledge evaluation, and model validation (see Chapters 12, 14.2, 16.1.3, 16.1.7, 19, and 34). The data mining engine contains a profiling module to segment an audience employing automatic decision trees, and predictive modeling involving discrete and continuous regression models and AI-based models to predict customers' choice. Numerous statistical tests are used to evaluate knowledge and validate results. GainSmarts' most unique component is feature selection. Governed by a set of rules, the process automatically selects the best predictors explaining customers' choice. GainSmarts also provides for scoring customer lists and using economic criteria to select customers for promotion. Detailed reporting and visualization tools facilitate understanding and interpretation of the model results. GainSmarts is designed for use either as a standalone or an open system. Migration of the system to additional applications is easy. Written primarily in SAS, GainSmarts can run on many SAS-supported platforms. GainSmarts has already been applied to a variety of problems in diverse industries. It is the two-time winner of the Gold Miner award in the KDD-CUP 97 and KDD-CUP 98 competitions, organized by the American Association of Artificial Intelligence.