Data Mining Model Structure
The structure of a data mining model is defined primarily by a set of data mining columns and a data mining algorithm. The data mining model content, created by the training process, is stored as data mining model nodes.
Each data mining column can contain one of several different content types, depending upon its use within the data mining model. Each column type has its own properties and behaviors. For more information, see Data Mining Columns.
The data mining algorithm uses the data mining column definitions to generate a predictive model by running the algorithm on training data submitted to the data mining model. The data mining model then stores the results obtained from analyzing the training data. Even though large amounts of training data may be inserted into a data mining model, the training data itself is not stored. Only the analysis information gained by processing that data and the distinct column values used as part of the analysis are stored as data mining model content.
Data Mining Model Nodes
Data mining model nodes represent the content of a data mining model. Each node contains information about the attributes needed to define the node, the relevant rules and other information needed to process a case against the node, and the analysis gained from training the node. Each node can also be related to other nodes, to support the complexities of decision tree and clustering algorithms in a common structure. The data mining model nodes can be browsed to further understand the decisions or aggregations made by the algorithm employed, and they can be modified to further adjust the data mining model.