odps/static/algorithms/classifier.xml

<?xml version='1.0' encoding='UTF-8'?> <algorithms baseClass="BaseTrainingAlgorithm"> <algorithm codeName="KNN"> <reloadFields>false</reloadFields> <baseClass>BaseProcessAlgorithm</baseClass> <docs><![CDATA[ KNN (k-Nearest Neighbor) algorithm classifies an instance by selecting K nearest neighbors and choose the label shared by largest numbers of nodes. Click `here <http://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm>`_ to get more information. Note that different to other algorithms, KNN use ``transform`` to predict and does not produce models. %params% :Example: >>> labeled = ds.label_field('class') >>> predicted = KNN(k=5).transform(labeled, to_be_predict) >>> predicted.store_odps('predicted') ]]> </docs> <params> <param name="trainTableName"> <exporter>get_input_table_name</exporter> <inputName>train</inputName> </param> <param name="trainFeatureColNames"> <exporter>get_feature_columns</exporter> <inputName>train</inputName> </param> <param name="trainLabelColName"> <exporter>get_label_column</exporter> <inputName>train</inputName> </param> <param name="trainTablePartitions"> <exporter>get_input_partitions</exporter> <inputName>train</inputName> </param> <param name="predictTableName"> <exporter>get_input_table_name</exporter> <inputName>predict</inputName> </param> <param name="predictFeatureColNames"> <exporter>get_feature_columns</exporter> <inputName>predict</inputName> </param> <param name="predictTablePartitions"> <exporter>get_input_partitions</exporter> <inputName>predict</inputName> </param> <param name="appendColNames"> <exporter>get_original_columns</exporter> <inputName>predict</inputName> </param> <param name="outputTableName"> <exporter>get_output_table_name</exporter> <outputName>output</outputName> </param> <param name="outputTablePartition"> <exporter>get_output_table_partitions</exporter> <outputName>output</outputName> </param> <param name="k"> <value>100</value> <docs>size of nearest neighbors</docs> </param> </params> <ports> <port name="train"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</docs> </port> <port name="predict"> <ioType>INPUT</ioType> <sequence>2</sequence> <type>DATA</type> <docs>test set with the same number of features as train set</docs> </port> <port name="output"> <ioType>OUTPUT</ioType> <sequence>1</sequence> <type>DATA</type> <schema> <copyInput>predict</copyInput> <schema>predict_label: bigint</schema> </schema> <docs>test set with label information stored in ``predict_label`` field</docs> </port> </ports> <metas> <meta name="xflowName" value="knn"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="xgboost"> <docs><![CDATA[ xgboost是一种高效率boosting算法，适用回归和二分类问题，详见 https://github.com/dmlc/xgboost 。 PAI平台目前只开放树形模型（gbtree），暂不开放线性模型（gblinear）。 %params% ]]></docs> <reloadFields>false</reloadFields> <params> <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="labelColName"> <alias>labelCol</alias> <exporter>get_label_column</exporter> <inputName>input</inputName> <docs>输入表中标签列的列名</docs> </param> <param name="booster"> <value>gbtree</value> </param> <param name="inputTableName"> <exporter>get_input_table_name</exporter> <inputName>input</inputName> </param> <param name="inputPartitions"> <exporter>get_input_partitions</exporter> <inputName>input</inputName> </param> <param name="modelName"> <exporter>generate_model_name</exporter> <outputName>output</outputName> <docs>输出的模型名</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> <param name="silent"> <value>1</value> </param> <param name="eta"> <value>0.3</value> <min>0</min> <max>1</max> <docs>为了防止过拟合，更新过程中用到的收缩步长。在每次提升计算之后，算法会直接获得新特征的权重。 eta通过缩减特征的权重使提升计算过程更加保守</docs> </param> <param name="gamma"> <value>0</value> <min>0</min> <docs>最小损失衰减</docs> </param> <param name="max_depth"> <value>6</value> <min>0</min> <docs>树的最大深度</docs> </param> <param name="min_child_weight"> <value>1</value> <min>0</min> <docs>孩子节点中最小的样本权重和。如果一个叶子节点的样本权重和小于min_child_weight则拆分过程结束。在线性回归模型中，这个参数是指建立每个模型所需的最小样本数</docs> </param> <param name="max_delta_step"> <value>0</value> <min>0</min> <docs>每个树所允许的最大delta步进</docs> </param> <param name="colsample_bytree"> <value>1</value> <min>0</min> <max>1</max> <docs>在建立树时对特征采样的比例</docs> </param> <param name="base_score"> <value>0.5</value> <docs>初始的predict score</docs> </param> <param name="seed"> <value>0</value> <docs>随机数的种子</docs> </param> <param name="objective"> <value>binary:logistic</value> <docs>定义学习任务及相应的学习目标，可选的目标函数</docs> </param> <param name="eval_metric"> <value>error</value> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="xgboost"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="RandomForests"> <reloadFields>false</reloadFields> <docs><![CDATA[ Random forest is a classifier which contains multiple decision trees and its output category depends on the mode of individual tree output categories. %params% :Example: >>> labeled = ds.label_field('class') >>> model = RandomForests(tree_num=10).train(labeled) >>> predicted = model.predict(dataset_to_be_predicted) >>> predicted.store_odps('predicted') ]]> </docs> <params> <param name="inputTableName" required="true"> <exporter>get_input_table_name</exporter> <inputName>input</inputName> </param> <param name="featureColNames" required="true"> <alias>featureCols</alias> <exporter>get_feature_columns</exporter> <inputName>input</inputName> <docs>input columns. the feature column names selected from input table, which are used for training </docs> </param> <param name="isFeatureContinuous" required="true"> <exporter>get_feature_continuous</exporter> <inputName>input</inputName> <docs>types of feature columns, '1' indicates that features are continuous; '0' means features are discrete. '1,0,0' indicates that the first column is continuous and the second and third column are discrete in three feature columns. The number of values corresponds to the length of features. It is preferable to use `df1 = df.continuous(['field1', 'field2']) instead of setting this value directly. </docs> </param> <param name="labelColName" required="true"> <alias>labelCol</alias> <exporter>get_label_column</exporter> <inputName>input</inputName> <docs>label column.the name of label column selected from input table</docs> </param> <param name="modelName" required="true"> <exporter>generate_model_name</exporter> <outputName>output</outputName> <docs>the random forest model</docs> </param> <param name="inputTablePartitions"> <exporter>get_input_partitions</exporter> <inputName>input</inputName> </param> <param name="treeNum"> <value>10</value> <required>true</required> <min>0</min> <max>1000</max> <docs>the number of trees in forest and the range is (0, 1000]</docs> </param> <param name="weightColName"> <alias>weightCol</alias> <exporter>get_weight_column</exporter> <inputName>input</inputName> <docs>name of weight column in input table. If there is no weight column, then the parameter is None. weightColName>0。 </docs> </param> <param name="algorithmTypes"> <exporter>$package_root.classifiers._customize.random_forest_algo_types</exporter> </param> <param name="randomColNum"> <inputName>input</inputName> <value>-1</value> <docs>num. of features: the feature number selected randomly when a single tree is generated. -1 indicates log2N(N refers to the total number of features) and effective value is [1,N] </docs> </param> <param name="minNumObj"> <value>2</value> <docs>minimal leaf size, the minimum number is 2</docs> </param> <param name="minNumPer"> <exporter>$package_root.classifiers._customize.random_forest_min_num_per</exporter> <docs>the minimal proportion accounted by leaf node number for parent nodes. -1 indicates no this limitation. The default value is -1 and the range is [0.0,1.0] </docs> </param> <param name="maxTreeDeep"> <value>10</value> <min>1</min> <docs>maximal depth of tree: -1 indicates completed growth. The range is [1, ∞)</docs> </param> <param name="maxRecordSize"> <value>100000</value> <min>1000</min> <max>1000000</max> <docs>maximal num. of instance per tree: the random data number of single tree in forest. The range is (1000, 1000000]. -1 indicates that the record number is 100000 </docs> </param> <param name="algorithmType"> <value>mix</value> <exported>false</exported> <docs>the algorithm types to be selected include: 'id3', 'cart', 'c45' and the default type 'mix', which divides three algorithms mentioned above equally </docs> </param> <param name="randomAttrType"> <value>logN</value> <exported>false</exported> <inputName>input</inputName> <docs>the feature number selected randomly when a single tree is generated. The types to be selected include 'logN', 'N/3', 'sqrtN' and 'N' </docs> </param> <param name="minNumPerInt"> <min>0</min> <max>100</max> <exported>false</exported> <docs>the random data number of single tree in forest. The range is (1000, 1000000]</docs> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="RandomForests"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="LogisticRegression"> <reloadFields>false</reloadFields> <docs><![CDATA[ Logistic regression is a classification method utilizing logistic functions. Detailed information can be found `here <https://en.wikipedia.org/wiki/Logistic_regression>`_. Note that although the original logistic regression algorithm only support binary classification, algorithm implemented in ODPS ML extends it using one-vs-rest strategy. It also use L-BFGS to improve performance. %params% :Example: >>> labeled = ds.label_field('class') >>> model = LogisticRegression(max_iter=100).train(labeled) >>> predicted = model.predict(dataset_to_be_predicted) >>> predicted.store_odps('predicted') ]]> </docs> <params> <param name="inputTableName" required="true"> <exporter>get_input_table_name</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, selected from input table and used for training</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>generate_model_name</exporter> <outputName>output</outputName> <docs>name of output model</docs> </param> <param name="inputTablePartitions"> <exporter>get_input_partitions</exporter> <inputName>input</inputName> </param> <param name="maxIter"> <value>100</value> <min>1</min> <max>10000</max> <docs>maximual iterations of L-BFGS. The default value is 100</docs> </param> <param name="epsilon"> <value>0.000001</value> <max>1000</max> <docs>termination condition of L-BFGS, that is the log-likelihood deviation between two interations. The default is 1.0e-06 </docs> </param> <param name="regularizedType"> <value>l1</value> <max>1000</max> <docs>regularization type. You can select 'l1', 'l2' or None and the default is 'l1'</docs> </param> <param name="goodValue"> <max>1000</max> <docs>positive label. If you select binary classification, specify the label value of training coefficient. If it is null, system will select a value randomly </docs> </param> <param name="regularizedLevel"> <value>1</value> <docs>regularization coeffcient: default value is 1.0; When regularizedType is 'None', this item will be ignored </docs> </param> <param name="twoClassification"> <value>false</value> <exported>false</exported> <docs>default is multionmial logistic regression. If this item is checked, it indicates selecting binary classification </docs> </param> <param name="enableSparse"> <exporter>get_enable_sparse</exporter> <inputName>input</inputName> </param> <param name="kvDelimiter"> <exporter>get_kv_delimiter</exporter> <inputName>input</inputName> </param> <param name="itemDelimiter"> <exporter>get_item_delimiter</exporter> <inputName>input</inputName> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="LogisticRegression"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="LinearSVM"> <docs><![CDATA[ Support Vector Machine (SVM) is a kind of pattern recognition method based on VC dimension theory and minimal structural risk principle. This version only supports linear binary classification. %params% :Example: >>> labeled = ds.label_field('class') >>> model = LinearSVM(epsilon=0.001).train(labeled) >>> predicted = model.predict(dataset_to_be_predicted) >>> predicted.store_odps('predicted') ]]></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</docs> </param> <param name="modelName" required="true"> <exporter>generate_model_name</exporter> <outputName>output</outputName> <docs>name of output model</docs> </param> <param name="featureColNames" required="true"> <alias>featureCols</alias> <exporter>get_feature_columns</exporter> <inputName>input</inputName> <docs>names of feature columns selected from input table, used for training</docs> </param> <param name="positiveLabel"> <docs>positive label. If it is not specified, select a random value. When there is an obvious quantity variance between the positive and negative samples, you are supported to specify a value as the positive label </docs> </param> <param name="epsilon"> <value>0.001</value> <min>0</min> <max>1</max> <docs>convergence coefficient: convergence error, the default value is 0.001 and the range is (0, 1) </docs> </param> <param name="positiveCost"> <value>1</value> <docs>positive weight: the default value is 1.0 and its range is (0, ~)</docs> </param> <param name="negativeCost"> <value>1</value> <docs>negative weight: the default value is 1.0 and its range is (0, ~)</docs> </param> <param name="cost"> <value>1</value> <exported>false</exported> <setter>$package_root.classifiers._customize.linear_svm_set_cost</setter> <docs>shortcut for setting both positive cost and negative cost</docs> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="LinearSVM"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="GBDT_LR"> <docs><![CDATA[ Gradient Boosting Decision Tree is a machine learning technique for regression problems, which produces a prediction model in the form of an ensemble of weak prediction models, typically decision trees. It builds the model in a stage-wise fashion like other boosting methods do, and it generalizes them by allowing optimization of an arbitrary differentiable loss function. This GBDT_LR module is used for binary classification. That is to set a threshold, if the label value is greater than the threshold, it is indicated as a positive example, otherwise it is indicated as a negative example. %params% :Example: >>> labeled = ds.label_field('class') >>> model = GBDTLR(tree_count=500).train(labeled) >>> predicted = model.predict(dataset_to_be_predicted) >>> predicted.store_odps('predicted') ]]></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="modelName" required="true"> <exporter>generate_model_name</exporter> <outputName>output</outputName> <docs>name of modle to be output</docs> </param> <param name="labelColName" required="true"> <alias>labelCol</alias> <exporter>get_label_column</exporter> <inputName>input</inputName> <docs>name of label column selected from input table</docs> </param> <param name="featureColNames" required="true"> <alias>featureCols</alias> <exporter>get_feature_columns</exporter> <inputName>input</inputName> <docs>feature columns of input table, used for training</docs> </param> <param name="groupIDColName"> <exporter>get_group_id_column</exporter> <inputName>input</inputName> <docs>name of grouping column. The default is to take a whole table as a group</docs> </param> <param name="metricType"> <value>2</value> <min>1</min> <max>10000</max> </param> <param name="treeCount"> <value>500</value> <min>1</min> <max>10000</max> <docs>num. of trees: the range is [1,10000] and default value is 500</docs> </param> <param name="shrinkage"> <value>0.05</value> <min>0</min> <max>1</max> <docs>leaning rate: the range is (0,1] and default value is 0.05</docs> </param> <param name="maxLeafCount"> <value>32</value> <min>2</min> <max>1000</max> <docs>maximal num. of leaves: it must be an integer. The range is [2,1000] and the default value is 32 </docs> </param> <param name="maxDepth"> <value>11</value> <min>1</min> <max>11</max> <docs>maximal depth of tree: it must be an integer. The range is [1,11] and the default value is 11 </docs> </param> <param name="minLeafSampleCount"> <value>500</value> <min>100</min> <max>1000</max> <docs>minimal leaf size: it must be an integer and the range is [100,1000]. The default value is 500 </docs> </param> <param name="sampleRatio"> <value>0.6</value> <min>0</min> <max>1</max> <docs>ratio of instances for training: the range is (0,1] and the default value is 0.6</docs> </param> <param name="featureRatio"> <value>0.6</value> <min>0</min> <max>1</max> <docs>ratio of features for training: the range is (0,1] and the default value is 0.6</docs> </param> <param name="testRatio"> <value>0.0</value> <min>0</min> <max>1</max> <docs>ratio of test instances: the range is [0,1) and the default value is 0.0</docs> </param> <param name="randSeed"> <value>0</value> <min>0</min> <max>10</max> <docs>random Seed: it must be an integer and the range is [0,10]. The default value is 0</docs> </param> <param name="featureSplitValueMaxSize"> <value>500</value> <min>1</min> <max>1000</max> <docs>maximal splits: the range is [1,1000] and the default value is 500</docs> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="GBDT_LR"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="NaiveBayes"> <reloadFields>false</reloadFields> <docs><![CDATA[ Naive Bayes classifiers are a family of simple probabilistic classifiers based on applying Bayes' theorem with strong (naive) independence assumptions between the features in machine learning. Click `here <http://en.wikipedia.org/wiki/Naive_Bayes_classifier>`_ for more information. %params% :Example: >>> labeled = ds.label_field('class') >>> model = NaiveBayes().train(labeled) >>> predicted = model.predict(dataset_to_be_predicted) >>> predicted.store_odps('predicted') ]]></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="featureColNames" required="true"> <alias>featureCols</alias> <exporter>get_feature_columns</exporter> <inputName>input</inputName> <docs>feature columns in input table, used for training</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>generate_model_name</exporter> <outputName>output</outputName> <docs>name of generated model</docs> </param> <param name="isFeatureContinuous" required="true"> <exporter>get_feature_continuous</exporter> <inputName>input</inputName> <docs>whether the feature is continuous. 1: continuous; 0: discrete. '1,0,0' indicates that the first column is continuous and the second column and third column are discrete. The value number corresponds to the length of features </docs> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</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="NaiveBayes"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="GLM"> <public>false</public> <reloadFields>false</reloadFields> <params> <param name="inputTableName" required="true"> <exporter>get_input_table_name</exporter> <inputName>input</inputName> </param> <param name="inputPartitions"> <exporter>get_input_partitions</exporter> <inputName>input</inputName> </param> <param name="featureColNames" required="true"> <alias>featureCols</alias> <exporter>get_feature_columns</exporter> <inputName>input</inputName> <docs>feature columns in input table, used for training</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="outputModelTableName" required="true"> <exporter>get_output_model_table_name(table_name=model)</exporter> <outputName>output</outputName> </param> <param name="distribute"> <value>gamma</value> <docs>数据分布</docs> </param> <param name="varPower"> <value>0</value> <docs>变量的指数，只有分布为Tweedie才有效</docs> </param> <param name="link"> <value>log</value> <docs>线性变换函数</docs> </param> <param name="linkPower"> <value>1</value> <docs>指数函数的指数，只有link为Power才有效</docs> </param> <param name="maxIter"> <value>1</value> <docs>最大迭代次数</docs> </param> <param name="epsilon"> <value>0.000001</value> <docs>收敛精度</docs> </param> </params> <ports> <port name="input"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>DATA</type> <docs>train set with defined features and labels</docs> </port> <port name="output"> <ioType>OUTPUT</ioType> <sequence>1</sequence> <type>MODEL</type> <docs>trained model which can be used in prediction</docs> <model> <type>TablesModel</type> <copyParams>method</copyParams>      </model> </port> </ports> <metas> <meta name="xflowName" value="glm"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> <algorithm codeName="GLMPredict"> <public>false</public> <baseClass>BaseProcessAlgorithm</baseClass> <reloadFields>false</reloadFields> <params> <param name="inputTableName" required="true"> <exporter>get_input_table_name</exporter> <inputName>input</inputName> </param> <param name="inputPartitions"> <exporter>get_input_partitions</exporter> <inputName>input</inputName> </param> <param name="inputModelTableName" required="true"> <exporter>get_input_model_table_name(table_name=model)</exporter> <inputName>model</inputName> </param> <param name="outputTableName" required="true"> <exporter>get_output_table_name</exporter> <outputName>output</outputName> </param> <param name="appendColNames"> <exporter>get_original_columns</exporter> <inputName>input</inputName> </param> </params> <ports> <port name="model"> <ioType>INPUT</ioType> <sequence>1</sequence> <type>MODEL</type> </port> <port name="input"> <ioType>INPUT</ioType> <sequence>2</sequence> <type>DATA</type> </port> <port name="output"> <ioType>OUTPUT</ioType> <sequence>1</sequence> <type>DATA</type> <schema> <copyInput>input</copyInput> <schema>conclusion: bigint: predicted_class</schema> </schema> </port> </ports> <metas> <meta name="xflowName" value="glm_predict"/> <meta name="xflowProjectName" value="algo_public"/> </metas> </algorithm> </algorithms>

odps/static/algorithms/classifier.xml (683 lines of code) (raw):