Preparatory Work For Data Mining

Before you start creating models, you must
  1. Formulate the problem to be solved.
  2. Locate the relevant data.
  3. Move the data to an Oracle database, if necessary. You can use the Import transformation to load flat flies to an Oracle database.
  4. Prepare the data appropriately (select the attributes to use, treat missing values, treat outliers, possibly bin the data, etc.) Oracle Data Miner transformations can be used to perform many of these tasks.
  5. Create a table for building the model and a different one for testing the model (for classification models). This is often done by splitting the table containing the data. You can use the Split transformation to split a table.
  6. Select the algorithm to use for model building.

Preparatory work is required to obtain meaningful results from your data mining project. The transformation on the Data menu help you to create transformations split tables. The preparatory work must be completed before you build models.

Data to which you apply a model must be prepared in the same way that the data used to build model was.

An Activity Guide helps you perform the preparatory worked in an organized manner. Once you choose the algorithm to use, the activity guide guides you though any required data preparation and also the model build or apply.