odps/static/algorithms/feature.xml

<?xml version='1.0' encoding='UTF-8'?>
<algorithms baseClass="BaseTrainingAlgorithm">
    <algorithm codeName="RFImportance">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        计算随机森林模型中特征重要性。

        %params%
        ]]></docs>
        <params>
            <param name="inputTableName" required="true">
                <exporter>get_input_table_name</exporter>
                <inputName>input</inputName>
            </param>
            <param name="inputTablePartitions">
                <exporter>get_input_table_partitions</exporter>
                <inputName>input</inputName>
            </param>
            <param name="featureColNames" required="true">
                <alias>featureCols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <docs>names of feature columns</docs>
            </param>
            <param name="labelColName" required="true">
                <alias>labelCol</alias>
                <exporter>get_label_column</exporter>
                <inputName>input</inputName>
                <docs>name of label column in input table</docs>
            </param>
            <param name="modelName" required="true">
                <exporter>get_input_model_name</exporter>
                <inputName>model</inputName>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>train set with defined label</docs>
            </port>
            <port name="model">
                <ioType>INPUT</ioType>
                <sequence>2</sequence>
                <type>MODEL</type>
                <docs>random forests model</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>MODEL</type>
                <docs>trained model which can be used in prediction</docs>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="feature_importance"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.feature._customize.get_rf_importance"/>
        </metas>
    </algorithm>
    <algorithm codeName="GBDTImportance">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        计算 GBDT 模型中特征重要性。

        %params%
        ]]></docs>
        <params>
            <param name="inputTableName" required="true">
                <exporter>get_input_table_name</exporter>
                <inputName>input</inputName>
            </param>
            <param name="inputTablePartitions">
                <exporter>get_input_table_partitions</exporter>
                <inputName>input</inputName>
            </param>
            <param name="labelColName" required="true">
                <alias>labelCol</alias>
                <exporter>get_label_column</exporter>
                <inputName>input</inputName>
                <docs>name of label column in input table</docs>
            </param>
            <param name="modelName" required="true">
                <exporter>get_input_model_name</exporter>
                <inputName>model</inputName>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="featureColNames" required="true">
                <alias>featureCols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <docs>names of feature columns</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>train set with defined label</docs>
            </port>
            <port name="model">
                <ioType>INPUT</ioType>
                <sequence>2</sequence>
                <type>MODEL</type>
                <docs>GBDT model</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>trained model which can be used in prediction</docs>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="gbdt_importance"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.feature._customize.get_gbdt_importance"/>
        </metas>
    </algorithm>
    <algorithm codeName="RegressionImportance">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        计算线性和二分类回归模型中特征重要性。

        %params%
        ]]></docs>
        <params>
            <param name="inputTableName" required="true">
                <exporter>get_input_table_name</exporter>
                <inputName>input</inputName>
            </param>
            <param name="inputTablePartitions">
                <exporter>get_input_partitions</exporter>
                <inputName>input</inputName>
            </param>
            <param name="labelColName" required="true">
                <alias>labelCol</alias>
                <exporter>get_label_column</exporter>
                <inputName>input</inputName>
                <docs>name of label column in input table</docs>
            </param>
            <param name="modelName" required="true">
                <exporter>get_input_model_name</exporter>
                <inputName>model</inputName>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="featureColNames">
                <alias>featureCols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <docs>输入表中用于训练的特征的列名</docs>
            </param>
            <param name="itemDelimiter">
                <exporter>get_item_delimiter</exporter>
                <inputName>input</inputName>
                <docs>item(key value对)之间的分隔符</docs>
            </param>
            <param name="kvDelimiter">
                <exporter>get_kv_delimiter</exporter>
                <inputName>input</inputName>
                <docs>表中每个item的key和value之间的分隔符</docs>
            </param>
            <param name="enableSparse">
                <exporter>get_enable_sparse</exporter>
                <inputName>input</inputName>
                <docs>是否稀疏数据</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>train set with defined label</docs>
            </port>
            <port name="model">
                <ioType>INPUT</ioType>
                <sequence>2</sequence>
                <type>MODEL</type>
                <docs>regression model</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>MODEL</type>
                <docs>trained model which can be used in prediction</docs>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="regression_feature_importance"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.feature._customize.get_regression_importance"/>
        </metas>
    </algorithm>
    <algorithm codeName="SelectFeatures">
        <baseClass>BaseProcessAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <params>
            <param name="inputTable" required="true">
                <exporter>get_input_table_name</exporter>
                <inputName>input</inputName>
            </param>
            <param name="inputTablePartitions">
                <exporter>get_input_partitions</exporter>
                <inputName>input</inputName>
            </param>
            <param name="outputTable">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="featImportanceTable" required="true">
                <exporter>get_output_table_name</exporter>
                <outputName>importance</outputName>
            </param>
            <param name="labelCol" required="true">
                <exporter>get_label_column</exporter>
                <inputName>input</inputName>
                <docs>标签列/目标列</docs>
            </param>
            <param name="selectedCols" required="true">
                <alias>featureCols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <docs>特征列</docs>
            </param>
            <param name="categoryCols">
                <inputName>input</inputName>
                <docs>存在在Int或者Double字符的枚举特征</docs>
            </param>
            <param name="maxBins">
                <docs>连续类特征划分最大区间数</docs>
            </param>
            <param name="selectMethod">
                <value>iv</value>
                <docs>特征选择方法</docs>
            </param>
            <param name="topN">
                <value>10</value>
                <docs>挑选的TopN个特征，如果topN答应输入特征数，则输出所有特征</docs>
            </param>
            <param name="itemSpliter">
                <alias>itemDelimiter</alias>
                <exporter>get_item_delimiter</exporter>
                <inputName>input</inputName>
                <docs>item(key value对)之间的分隔符</docs>
            </param>
            <param name="kvSpliter">
                <alias>kvDelimiter</alias>
                <exporter>get_kv_delimiter</exporter>
                <inputName>input</inputName>
                <docs>表中每个item的key和value之间的分隔符</docs>
            </param>
            <param name="isSparse">
                <alias>enableSparse</alias>
                <exporter>get_enable_sparse</exporter>
                <inputName>input</inputName>
                <docs>是否稀疏数据</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>train set with input data</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>output table with selected features</docs>
                <schema>
                    <schema>{labelCol}</schema>
                    <dynamic>true</dynamic>
                </schema>
            </port>
            <port name="importance">
                <ioType>OUTPUT</ioType>
                <sequence>2</sequence>
                <type>DATA</type>
                <docs>importance table</docs>
                <schema>
                    <schema>featname: string, weight: double</schema>
                </schema>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="fe_select_runner"/>
            <meta name="xflowProjectName" value="algo_public"/>
        </metas>
    </algorithm>
</algorithms>