Configure the parameters of the GBDT Regression component - Platform For AI

Gradient Boosting Decision Tree (GBDT) is an advanced machine learning algorithm. It uses an iterative method to build a collection of decision trees for regression analysis. By progressively optimizing a loss function, the algorithm can handle both linear and nonlinear regression problems and provide highly accurate prediction results.

Configure the component

Method 1: Visual configuration

Add the GBDT Regression component to the Designer workflow page. Then, configure the parameters in the pane on the right.

Parameter type	Parameter	Description
Fields setting	Input columns	The feature columns from the input data source used for training. Supports DOUBLE and BIGINT types. Note The number of feature columns cannot exceed 800.
	Label column	Supports DOUBLE and BIGINT types.
	Group column	Supports DOUBLE and BIGINT types. By default, the entire table is treated as one group.
Parameters setting	Loss function type	Supported types are gbrank loss, lambdamart dcg loss, lambdamart ndcg loss, and regression loss.
	Tau parameter in gbrank loss	The value must be in the range of [0,1].
	Exponent base in gbrank and regression loss	The value must be in the range of [1,10].
	Metric type	Supported types are NDCG and DCG.
	Number of trees	The value must be in the range of 1 to 10000.
	Learning rate	The value must be in the range of (0,1).
	Maximum number of leaf nodes	The value must be in the range of 1 to 1000.
	Maximum depth of a tree	The value must be in the range of 1 to 11.
	Minimum number of samples on a leaf node	The value must be in the range of 1 to 1000.
	Sample sampling ratio	The value must be in the range of (0,1).
	Feature sampling ratio during training	The value must be in the range of (0,1).
	Ratio of test samples	The value must be in the range of [0,1).
	Seed for random number generator	The value must be in the range of [0,10].
	Use Newton's method for learning	Specifies whether to use Newton's method.
	Maximum number of splits for a feature	The value must be in the range of 1 to 1000.
Execution tuning	Number of computing cores	The system automatically allocates the number of training instances based on the input data volume.
Execution tuning	Memory per core	The system automatically allocates memory based on the input data volume. The unit is MB.

Method 2: Use PAI commands

Use PAI commands to configure the parameters for the GBDT Regression component. You can use the SQL Script component to call PAI commands. For more information, see SQL Script.

PAI -name gbdt
    -project algo_public
    -DfeatureSplitValueMaxSize="500"
    -DlossType="0"
    -DrandSeed="0"
    -DnewtonStep="0"
    -Dshrinkage="0.05"
    -DmaxLeafCount="32"
    -DlabelColName="campaign"
    -DinputTableName="bank_data_partition"
    -DminLeafSampleCount="500"
    -DsampleRatio="0.6"
    -DgroupIDColName="age"
    -DmaxDepth="11"
    -DmodelName="xlab_m_GBDT_83602"
    -DmetricType="2"
    -DfeatureRatio="0.6"
    -DinputTablePartitions="pt=20150501"
    -Dtau="0.6"
    -Dp="1"
    -DtestRatio="0.0"
    -DfeatureColNames="previous,cons_conf_idx,euribor3m"
    -DtreeCount="500"

Parameter	Required	Default value	Description
inputTableName	Yes	None	The name of the input table.
featureColNames	No	All numeric columns	The names of the feature columns in the input table used for training. Supports DOUBLE and BIGINT types.
labelColName	Yes	None	The name of the label column in the input table. Supports DOUBLE and BIGINT types.
inputTablePartitions	No	All partitions	The partitions in the input table used for training. The following formats are supported: Partition_name=value name1=value1/name2=value2: multi-level partitions Note Use commas (,) to separate multiple partitions.
modelName	Yes	None	The name of the output model.
outputImportanceTableName	No	None	The name of the output table for feature importance.
groupIDColName	No	Full table	The data grouping column.
lossType	No	0	The loss function. The following types are supported: 0: GBRANK 1: LAMBDAMART_DCG 2: LAMBDAMART_NDCG 3: LEAST_SQUARE
metricType	No	0	The types include the following: 0: NDCG (Normalized Discounted Cumulative Gain) 1: DCG (Discounted Cumulative Gain) 2: AUC. This type applies only when the label value is 0/1. This type is deprecated.
treeCount	No	500	The number of trees. The value must be in the range of 1 to 10000.
shrinkage	No	0.05	The learning rate. The value must be in the range of (0,1).
maxLeafCount	No	32	The maximum number of leaf nodes. The value must be in the range of 1 to 1000.
maxDepth	No	10	The maximum depth of a tree. The value must be in the range of 1 to 11.
minLeafSampleCount	No	500	The minimum number of samples on a leaf node. The value must be in the range of 1 to 1000.
sampleRatio	No	0.6	The sampling ratio for samples during training. The value must be in the range of (0,1).
featureRatio	No	0.6	The sampling ratio for features during training. The value must be in the range of (0,1).
tau	No	0.6	The Tau parameter in GBRank Loss. The value must be in the range of [0,1].
p	No	1	The p parameter in GBRank Loss. The value must be in the range of [1,10].
randSeed	No	0	The seed for the random number generator. The value must be in the range of [0,10].
newtonStep	No	1	Specifies whether to use the Newton iteration method. The value can be {0,1}.
featureSplitValueMaxSize	No	500	The maximum number of splits for a feature. The value must be in the range of 1 to 1000.
lifecycle	No	None	The lifecycle of the output table.

Example

Use an SQL statement to generate test data.

drop table if exists gbdt_ls_test_input;
create table gbdt_ls_test_input
as
select
    *
from
(
    select
        cast(1 as double) as f0,
        cast(0 as double) as f1,
        cast(0 as double) as f2,
        cast(0 as double) as f3,
        cast(0 as bigint) as label
    union all
        select
            cast(0 as double) as f0,
            cast(1 as double) as f1,
            cast(0 as double) as f2,
            cast(0 as double) as f3,
            cast(0 as bigint) as label
    union all
        select
            cast(0 as double) as f0,
            cast(0 as double) as f1,
            cast(1 as double) as f2,
            cast(0 as double) as f3,
            cast(1 as bigint) as label
    union all
        select
            cast(0 as double) as f0,
            cast(0 as double) as f1,
            cast(0 as double) as f2,
            cast(1 as double) as f3,
            cast(1 as bigint) as label
    union all
        select
            cast(1 as double) as f0,
            cast(0 as double) as f1,
            cast(0 as double) as f2,
            cast(0 as double) as f3,
            cast(0 as bigint) as label
    union all
        select
            cast(0 as double) as f0,
            cast(1 as double) as f1,
            cast(0 as double) as f2,
            cast(0 as double) as f3,
            cast(0 as bigint) as label
) a;

The generated test data table gbdt_ls_test_input is as follows.

f0	f1	f2	f3	label
1.0	0.0	0.0	0.0	0
0.0	0.0	1.0	0.0	1
0.0	0.0	0.0	1.0	1
0.0	1.0	0.0	0.0	0
1.0	0.0	0.0	0.0	0
0.0	1.0	0.0	0.0	0

Use a PAI command to submit the training parameters for the GBDT Regression component.

drop offlinemodel if exists gbdt_ls_test_model;
PAI -name gbdt
    -project algo_public
    -DfeatureSplitValueMaxSize="500"
    -DlossType="3"
    -DrandSeed="0"
    -DnewtonStep="1"
    -Dshrinkage="0.5"
    -DmaxLeafCount="32"
    -DlabelColName="label"
    -DinputTableName="gbdt_ls_test_input"
    -DminLeafSampleCount="1"
    -DsampleRatio="1"
    -DmaxDepth="10"
    -DmetricType="0"
    -DmodelName="gbdt_ls_test_model"
    -DfeatureRatio="1"
    -Dp="1"
    -Dtau="0.6"
    -DtestRatio="0"
    -DfeatureColNames="f0,f1,f2,f3"
    -DtreeCount="10"

Use a PAI command to submit the parameters for the Prediction component.

drop table if exists gbdt_ls_test_prediction_result;
PAI -name prediction
    -project algo_public
    -DdetailColName="prediction_detail"
    -DmodelName="gbdt_ls_test_model"
    -DitemDelimiter=","
    -DresultColName="prediction_result"
    -Dlifecycle="28"
    -DoutputTableName="gbdt_ls_test_prediction_result"
    -DscoreColName="prediction_score"
    -DkvDelimiter=":"
    -DinputTableName="gbdt_ls_test_input"
    -DenableSparse="false"
    -DappendColNames="label"

View the prediction result table gbdt_ls_test_prediction_result.

label	prediction_result	prediction_score	prediction_detail
0	NULL	0.0	{“label”: 0}
0	NULL	0.0	{“label”: 0}
1	NULL	0.9990234375	{“label”: 0.9990234375}
1	NULL	0.9990234375	{“label”: 0.9990234375}
0	NULL	0.0	{“label”: 0}
0	NULL	0.0	{“label”: 0}