DecisionForest Example 3: Regression Tree with Out-of-Bag Error - Teradata Vantage

This example creates a regression tree to predict the median value of homes in $1000s.

Input

The InputTable has Boston housing value data. Predictors include crime rate, proportion of nonretail business, average number of rooms in each dwelling, accessibility to radial highways, and property tax rate.

SQL Call

SELECT * FROM DecisionForest (
  ON boston AS InputTable
  OUT TABLE OutputTable (rtf_model)
  OUT TABLE MonitorTable(boston_monitor_table)
  USING
  ResponseColumn ('medv') 
  NumericInputs ('crim','zn','indus','chas','nox','rm',
    'age','dis','rad','tax','ptratio','black','lstat')
  TreeType ('regression')
  MinNodeSize ('2')
  MaxDepth ('6') 
  NumTrees ('50')
  OutOfBag ('true') 
) AS dt;

Output

Computing 48 regression trees.
Each worker is computing 16 trees.
Each tree will contain approximately 168 points.
Poisson sampling parameter: 0.996
Mean of squared residuals: 15.396538832527742
% Var explained: 81.76188132990643
Decision forest created.

This query returns the following table:

SELECT task_index, tree_num, cast (tree AS VARCHAR(50))
  FROM rft_model ORDER BY 1;