DocsICP FilingConsole
官方文档
首页

Random Forest

更新时间：
复制 MD 格式
产品详情
我的收藏

A Random Forest is a classifier containing multiple decision trees. Its output is the mode of the classes from the individual trees.

Component configuration

You can configure the parameters for the Random Forest component by using one of the following methods.

Method 1: GUI

Configure the component parameters on the pipeline page in Machine Learning Designer.

Tab

Parameter

Description

field settings

feature columns

By default, all columns are used except the label and weight columns.

excluded columns

This parameter cannot be used with feature columns.

Force conversion columns

The parsing rules are as follows:

  • Columns of the STRING, BOOLEAN, and DATETIME types are parsed as a discrete type.

  • Columns of the DOUBLE and BIGINT types are parsed as a continuous type.

Note

To parse a BIGINT column as CATEGORICAL, use the forceCategorical parameter.

weight column

The column used to weigh each sample row. Only numeric types are supported.

label column

The label column of the input table. STRING and numeric types are supported.

parameter settings

Number of trees

Valid values: 1 to 1000.

Algorithm distribution

Specifies how different tree algorithms are distributed in the forest. If a forest has N trees and you set this parameter to algorithmTypes=[a,b], the algorithms are allocated as follows:

  • The range [0,a) uses the ID3 algorithm.

  • The range [a,b) uses the CART algorithm.

  • The range [b,N] uses the C4.5 algorithm.

For example, in a forest with five trees, if you set the parameter to [2,4], tree 1 uses the ID3 algorithm, trees 2 and 3 use the CART algorithm, and trees 4 and 5 use the C4.5 algorithm. If you enter None, the algorithms are distributed evenly across the forest.

Number of random features per tree

Valid values: [1,N], where N is the total number of features.

Minimum samples per leaf node

Must be a positive integer. The default value is 2.

Minimum sample ratio per leaf node

The minimum ratio of samples in a leaf node relative to its parent node. Valid values: [0,1]. The default value is 0.

Maximum tree depth

Valid values: [1,+∞). The default value is unlimited.

Number of random input samples per tree

Valid values: (1000,1000000]. The default value is 100000.

Method 2: PAI command

Use a PAI command to configure the component parameters. You can run the PAI command by using the SQL script component. For more information, see SQL script.

 PAI -name randomforests
     -project algo_public
     -DinputTableName="pai_rf_test_input"
     -DmodelName="pai_rf_test_model"
     -DforceCategorical="f1"
     -DlabelColName="class"
     -DfeatureColNames="f0,f1"
     -DmaxRecordSize="100000"
     -DminNumPer="0"
     -DminNumObj="2"
     -DtreeNum="3";

Parameter

Required

Description

Default

inputTableName

Yes

The input table.

N/A

inputTablePartitions

No

The input table partitions for training. Supported formats:

  • Partition_name=value

  • name1=value1/name2=value2: a multi-level format

Note

To specify multiple partitions, separate them with commas (,).

All partitions

labelColName

Yes

The name of the label column in the input table.

N/A

modelName

Yes

The name of the output model.

N/A

treeNum

Yes

The number of trees in the forest. Valid values: 1 to 1000.

100

excludedColNames

No

This parameter cannot be used with featureColNames.

Empty

weightColName

No

The name of the weight column in the input table.

N/A

featureColNames

No

The names of the feature columns used for training.

All columns except those specified for labelColName and weightColName.

forceCategorical

No

The parsing rules are as follows:

  • Columns of the STRING, BOOLEAN, and DATETIME types are parsed as a discrete type.

  • Columns of the DOUBLE and BIGINT types are parsed as a continuous type.

Note

To parse a BIGINT column as CATEGORICAL, use the forceCategorical parameter.

INT is treated as a continuous type.

algorithmTypes

No

Specifies how different tree algorithms are distributed in the forest. If a forest has N trees and algorithmTypes=[a,b]:

  • The range [0,a) uses the ID3 algorithm.

  • The range [a,b) uses the CART algorithm.

  • The range [b,N] uses the C4.5 algorithm.

For example, in a forest with five trees, with a setting of [2,4], tree 1 uses the ID3 algorithm, trees 2 and 3 use the CART algorithm, and trees 4 and 5 use the C4.5 algorithm. If you enter None, the algorithms are distributed evenly across the forest.

The algorithms are evenly distributed.

randomColNum

No

The number of random features to consider at each split when building a decision tree. Valid values: [1,N], where N is the total number of features.

log2N

minNumObj

No

The minimum number of samples required in a leaf node. Must be a positive integer.

2

minNumPer

No

The minimum ratio of samples in a leaf node relative to its parent node. Valid values: [0,1].

0.0

maxTreeDeep

No

The maximum depth of a single tree. Valid values: [1,+∞).

Unlimited

maxRecordSize

No

The number of random samples used to build a single tree. Valid values: (1000,1000000].

100000

Example

  1. Use the following SQL statement to generate training data:

    create table pai_rf_test_input as
select * from
(
  select 1 as f0,2 as f1, "good" as class
  union all
  select 1 as f0,3 as f1, "good" as class
  union all
  select 1 as f0,4 as f1, "bad" as class
  union all
  select 0 as f0,3 as f1, "good" as class
  union all
  select 0 as f0,4 as f1, "bad" as class
)tmp;

  2. Use the following PAI command to submit the Random Forest component parameters:

    PAI -name randomforests
     -project algo_public
     -DinputTableName="pai_rf_test_input"
     -DmodelName="pai_rf_test_model"
     -DforceCategorical="f1"
     -DlabelColName="class"
     -DfeatureColNames="f0,f1"
     -DmaxRecordSize="100000"
     -DminNumPer="0"
     -DminNumObj="2"
     -DtreeNum="3";

  3. View the model in Predictive Model Markup Language (PMML).

    <?xml version="1.0" encoding="utf-8"?>
<PMML xmlns="http://www.dmg.org/PMML-4_2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="4.2" xsi:schemaLocation="http://www.dmg.org/PMML-4_2 http://www.dmg.org/v4-2/pmml-4-2.xsd">
  <Header copyright="Copyright (c) 2014, Alibaba Inc." description="">
    <Application name="ODPS/PMML" version="0.1.0"/>
    <Timestamp>Tue, 12 Jul 2016 07:04:48 GMT</Timestamp>
  </Header>
  <DataDictionary numberOfFields="2">
    <DataField name="f0" optype="continuous" dataType="integer"/>
    <DataField name="f1" optype="continuous" dataType="integer"/>
    <DataField name="class" optype="categorical" dataType="string">
      <Value value="bad"/>
      <Value value="good"/>
    </DataField>
  </DataDictionary>
  <MiningModel modelName="xlab_m_random_forests_1_75078_v0" functionName="classification" algorithmName="RandomForests"/>
    <MiningSchema>
      <MiningField name="f0" usageType="active"/>
      <MiningField name="f1" usageType="active"/>
      <MiningField name="class" usageType="target"/>
    </MiningSchema>
    <Segmentation multipleModelMethod="majorityVote">
      <Segment id="0">
        <True/>
        <TreeModel modelName="xlab_m_random_forests_1_75078_v0" functionName="classification" algorithmName="RandomForests">
          <MiningSchema>
            <MiningField name="f0" usageType="active"/>
            <MiningField name="f1" usageType="active"/>
            <MiningField name="class" usageType="target"/>
          </MiningSchema>
          <Node id="1">
            <True/>
            <ScoreDistribution value="bad" recordCount="2"/>
            <ScoreDistribution value="good" recordCount="3"/>
            <Node id="2" score="good">
              <SimplePredicate field="f1" operator="equal" value="2"/>
              <ScoreDistribution value="good" recordCount="1"/>
            </Node>
            <Node id="3" score="good">
              <SimplePredicate field="f1" operator="equal" value="3"/>
              <ScoreDistribution value="good" recordCount="2"/>
            </Node>
            <Node id="4" score="bad"
              <SimplePredicate field="f1" operator="equal" value="4"/>
              <ScoreDistribution value="bad" recordCount="2"/>
            </Node>
          </Node>
        </TreeModel>
      </Segment>
      <Segment id="1">
        <True/>
        <TreeModel modelName="xlab_m_random_forests_1_75078_v0" functionName="classification" algorithmName="RandomForests">
          <MiningSchema>
            <MiningField name="f0" usageType="active"/>
            <MiningField name="f1" usageType="active"/>
            <MiningField name="class" usageType="target"/>
          </MiningSchema>
          <Node id="1">
            <True/>
            <ScoreDistribution value="bad" recordCount="2"/>
            <ScoreDistribution value="good" recordCount="3"/>
            <Node id="2" score="good">
              <SimpleSetPredicate field="f1" booleanOperator="isIn">
                <Array n="2" type="integer"2 3</Array>
              </SimpleSetPredicate>
              <ScoreDistribution value="good" recordCount="3"/>
            </Node>
            <Node id="3" score="bad">
              <SimpleSetPredicate field="f1" booleanOperator="isNotIn">
                <Array n="2" type="integer"2 3</Array>
              </SimpleSetPredicate>
              <ScoreDistribution value="bad" recordCount="2"/>
            </Node>
          </Node>
        </TreeModel>
      </Segment>
      <Segment id="2">
        <True/>
        <TreeModel modelName="xlab_m_random_forests_1_75078_v0" functionName="classification" algorithmName="RandomForests">
          <MiningSchema>
            <MiningField name="f0" usageType="active"/>
            <MiningField name="f1" usageType="active"/>
            <MiningField name="class" usageType="target"/>
          </MiningSchema>
          <Node id="1">
            <True/>
            <ScoreDistribution value="bad" recordCount="2"/>
            <ScoreDistribution value="good" recordCount="3"/>
            <Node id="2" score="bad">
              <SimplePredicate field="f0" operator="lessOrEqual" value="0.5"/>
              <ScoreDistribution value="bad" recordCount="1"/>
              <ScoreDistribution value="good" recordCount="1"/>
            </Node>
            <Node id="3" score="good">
              <SimplePredicate field="f0" operator="greaterThan" value="0.5"/>
              <ScoreDistribution value="bad" recordCount="1"/>
              <ScoreDistribution value="good" recordCount="2"/>
            </Node>
          </Node>
        </TreeModel>
      </Segment>
    </Segmentation>
  </MiningModel>
</PMML>

  4. View the visualized output of the model.

Previous：Logistic regression for multiclass classificationNext：Naive Bayes