tree depth - Platform For AI - Alibaba Cloud Documentation Center

Tree depth is defined as the number of nodes along the longest path from the root node to the farthest leaf node in a decision tree model. It is a critical hyperparameter that influences the model's complexity and ability to fit data. Deeper trees can capture more complex data patterns but may lead to overfitting, while shallower trees risk underfitting. Thus, choosing the right tree depth is essential for optimal model performance and generalization.

Configure widget

Method 1: Visualization method

On the Designer workflow page, add the Tree Depth widget and set the relevant parameters on the right side of the interface:

Parameter type	Parameter	Description
Field set	Starting column of the input edge list	The starting column of the edge list.
Field set	Ending column of the input edge list	The ending column of the edge list.
Execution optimization	Number of processes	The number of nodes for parallel execution of the job. The larger the number, the higher the degree of parallelism. However, the frame communication overhead will increase.
	Process memory	The maximum amount of memory available for a single job, in MB. The default value is 4096. If the actual memory usage exceeds this value, an `OutOfMemory` exception will be thrown.
	Data split size	The size of the data split, in MB. The default value is 64.

Method 2: PAI command method

To configure the Tree Depth widget parameters, use PAI commands. Employ the SQL script widget to execute PAI commands. For more information, see Scenario 4: Execute PAI Commands in the SQL Script Widget.

PAI -name TreeDepth
    -project algo_public
    -DinputEdgeTableName=TreeDepth_func_test_edge
    -DfromVertexCol=flow_out_id
    -DtoVertexCol=flow_in_id
    -DoutputTableName=TreeDepth_func_test_result;

Parameter	Required	Default value	Description
inputEdgeTableName	Yes	None	Name of the input edge list.
inputEdgeTablePartitions	No	Read the entire table	Partition of the input edge list.
fromVertexCol	Yes	None	The starting column of the input edge list.
toVertexCol	Yes	None	The ending column of the input edge list.
outputTableName	Yes	None	Name of the output table.
outputTablePartitions	No	None	Partition of the output table.
lifecycle	No	None	Lifecycle of the output table.
workerNum	No	Not set	The number of nodes for parallel execution of the job. The larger the number, the higher the degree of parallelism. However, the frame communication overhead will increase.
workerMem	No	4096	The maximum amount of memory available for a single job, in MB. The default value is 4096. If the actual memory usage exceeds this value, an `OutOfMemory` exception will be thrown.
splitSize	No	64	Data split size.

Usage example

Add an SQL script widget, deselect Use Script Mode and Add Create Table Statement By System, and input the following SQL statements in the SQL Script.

drop table if exists TreeDepth_func_test_edge;
create table TreeDepth_func_test_edge as
select * from
(
    select '0' as flow_out_id, '1' as flow_in_id
    union all
    select '0' as flow_out_id, '2' as flow_in_id
    union all
    select '1' as flow_out_id, '3' as flow_in_id
    union all
    select '1' as flow_out_id, '4' as flow_in_id
    union all
    select '2' as flow_out_id, '4' as flow_in_id
    union all
    select '2' as flow_out_id, '5' as flow_in_id
    union all
    select '4' as flow_out_id, '6' as flow_in_id
    union all
    select 'a' as flow_out_id, 'b' as flow_in_id
    union all
    select 'a' as flow_out_id, 'c' as flow_in_id
    union all
    select 'c' as flow_out_id, 'd' as flow_in_id
    union all
    select 'c' as flow_out_id, 'e' as flow_in_id
)tmp;
drop table if exists TreeDepth_func_test_result;
create table TreeDepth_func_test_result
(
  node string,
  root string,
  depth bigint
);

The corresponding data structure graph:

图结构

Add another SQL script widget, uncheck Use Script Mode and Add Create Table Statement By System, enter the PAI command into the SQL Script, and link the widgets from Steps 1 and 2.

drop table if exists ${o1};
PAI -name TreeDepth
    -project algo_public
    -DinputEdgeTableName=TreeDepth_func_test_edge
    -DfromVertexCol=flow_out_id
    -DtoVertexCol=flow_in_id
    -DoutputTableName=${o1};

Click the top left , and execute the workflow.

Upon completion, right-click the widget from Step 2, choose View Data > SQL Script Output, and examine the training results.

| node | root | depth |
| ---- | ---- | ----- |
| a    | a    | 0     |
| b    | a    | 1     |
| c    | a    | 1     |
| d    | a    | 2     |
| e    | a    | 2     |
| 0    | 0    | 0     |
| 1    | 0    | 1     |
| 2    | 0    | 1     |
| 3    | 0    | 2     |
| 4    | 0    | 2     |
| 5    | 0    | 2     |
| 6    | 0    | 3     |