<model name>_COLUMNS table
This table contains all columns that are used by the data mining algorithm.
The table might contain columns from the original user input table and columns that are internally created by the auto-transform option. For model application, only the original columns are required because the transformed values are determined by the scoring algorithm. Parameter values for the transformations, such as the mean value of a column, are contained in <model name>_COLUMN_STATISTICS. For some mining functions, only a subset of the usage types is possible.
The table contains one line for each model column.
The following table shows the table columns:
| Column name | Data type | Description |
|---|---|---|
| COLUMNNAME (primary key column) | VARCHAR(128) | Name of an input column |
| DATATYPE | VARCHAR(64) | SQL data type of the column |
| OPTYPE | VARCHAR(16) | Operational type of the column Possible values are categorical and continuous. |
| USAGETYPE | VARCHAR(16) | Usage type active |
| COLUMNWEIGHT | DOUBLE | A priori factor of contribution to the model training process relative to
other columns The default value is 1.0. |
| COMPAREFUNCTION | VARCHAR(16) | Function to compare a pair of column values Possible values are absDiff and equal. |
| IMPORTANCE | DOUBLE | Indicates the importance of the column for the data mining model as determined
by the algorithm The value is > 0. |
| OUTLIERTREATMENT | VARCHAR(16) | Outlier treatment The value is asIs. |
| LOWERLIMIT | DOUBLE | Lower limit of valid values in a numeric field Null indicates negative infinity. |
| UPPERLIMIT | DOUBLE | Upper limit of valid values in a numeric field Null indicates positive infinity. |
| CLOSURE | VARCHAR(12) | Indicates whether the outlier limits are contained in the valid value range
Possible values are openClosed, openOpen, closedOpen, and closedClosed. |
| STATISTICSTYPE | VARCHAR(16) | Indicates which type of statistics are available for the column Possible values are as follows:
|