Data Mining

The first step in developing a predictive model is to generate a dataset, which contains all the data to be used for predictions including any data to be predicted. The dataset is a MicroStrategy report, and usually has the following features:

• Each column represents a particular type of data that either has predictive value or represents an outcome worth predicting.

• Each row represents a specific business attribute, such as a customer or product.

The dataset report can be easily created, refreshed, and accessed, even if it is large and complex. Using the MicroStrategy data mart feature, this dataset report can be saved as a table in the database. It can also be exported into a number of file formats. Third party data mining applications can access the report as a data mart or as an exported file to develop predictive models.

Using a data mart has an added benefit of allowing the third-party data mining application to determine the data type of each column (that is, each predictive variable). Data types identify the way data should be interpreted, for example, as a string, an integer, or a floating point number. Exporting the dataset, for example into a text-only format, often causes this information to be lost. Using a data mart will help ensure the predictive model is developed with the proper data types for all variables.

After you have created the data mart, the next step is to create the predictive model.