New Decision Mining Tree
This Build tool is used to create a Decision Mining Object that uses an input Data Table to generate a probabilistic Decision or a Regression Tree.
The developer selects an exiting table (from a data mining server) and then selects the required data columns. An optional SQL WHERE clause allows a subset of the data rows to be selected. One use of this "WHERE clause" is if data sampling was used with Upload New Data option to create random training and test samples.
To select the a sample you can enter
"__sample=1"
This would only utilise rows that have their "sample" column set to 1 to generate Data Mining Tree from the data table.
Once a table has been selected you can map the columns to attributes (that can then also be used in other trees). You can also set the number of decimal places to be displayed for Numeric type columns. You can also decide if the column is included or excluded or should be the outcome of the tree. A single outcome column must be selected be used per tree.
You can then choose the Base Class for each column that you want to map to an attribute and also the Category under which each Attribute will be placed.
The data mining object is identified by this icon: 