odps/static/algorithms/statistics.xml

<?xml version='1.0' encoding='UTF-8'?>
<algorithms baseClass="BaseProcessAlgorithm">
    <algorithm codeName="Histograms">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
Output histograms for multiple columns.

%params%

:Returns: histograms for every columns, containing histograms and bins.

:Example:
>>> from odps.ml.statistics import *
>>> hists = histograms(ionosphere)
>>> # get a column from results
>>> hist, bins = hists['a04']
>>> # plot the result
>>> plt.bar(bins[:-1], hist, width=bins[1] - bins[0])
        ]]></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="outputTableName" required="true">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="selectedColNames" required="true">
                <alias>cols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <docs>selected columns, feature columns by default</docs>
            </param>
            <param name="intervalNum">
                <value>10</value>
                <docs>number of intervals for histograms</docs>
            </param>
            <param name="coreNum">
                <docs>number of cores for calculation</docs>
            </param>
            <param name="memSizePerCore">
                <docs>memory size per core</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data set</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="histograms"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_histograms"/>
        </metas>
    </algorithm>
    <algorithm codeName="Pearson">
        <docs><![CDATA[
Calculate Pearson coefficient for two fields in the input data set. If not specified, the first column will be label
column, while the second will be regressed value.

%params%

:Returns: pearson coefficient for two given columns.
]]></docs>
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <params>
            <param name="inputTableName">
                <exporter>get_input_table_name</exporter>
                <inputName>input</inputName>
            </param>
            <param name="inputTablePartitions">
                <exporter>get_input_partitions</exporter>
                <inputName>input</inputName>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="col1Name" required="true">
                <alias>col1</alias>
                <exporter>get_label_column</exporter>
                <inputName>input</inputName>
                <docs>the first selected column</docs>
            </param>
            <param name="col2Name">
                <alias>col2</alias>
                <required>true</required>
                <exporter>$package_root.statistics._customize.get_predicted_value_column</exporter>
                <inputName>input</inputName>
                <docs>the second selected column</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data set</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="pearson"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_pearson_result"/>
        </metas>
    </algorithm>
    <algorithm codeName="TTest">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
Student's T test verifies the equality of means when the sample conforms to normal distribution. This module supports
three types of T tests.

1. Test between a sample field and a fixed value. x_col and mu should be configured. e.g.,

>>> t_test(ds, x_col='col_name', mu=1)

2. Test between two independent normal-distributed samples. x_col and y_col as well as mu should be configured. paired should be False. e.g.,

>>> t_test(ds, x_col='col1', y_col='col2', mu=1, paired=False)

3. Test between two paired samples. paird should be True. e.g.,

>>> t_test(ds, x_col='col1', y_col='col2', mu=1, paired=True)

%params%

:Returns: a json object representing test report.
        ]]></docs>
        <params>
            <param name="xTableName">
                <exporter>get_input_table_name</exporter>
                <inputName>x</inputName>
            </param>
            <param name="xTablePartitions">
                <exporter>get_input_partitions</exporter>
                <inputName>x</inputName>
            </param>
            <param name="xColName">
                <alias>xCol</alias>
                <docs>column X</docs>
            </param>
            <param name="yTableName">
                <exporter>$package_root.statistics._customize.get_y_table_name</exporter>
                <inputName>y</inputName>
            </param>
            <param name="yTablePartitions">
                <exporter>$package_root.statistics._customize.get_y_partitions</exporter>
                <inputName>y</inputName>
            </param>
            <param name="yColName">
                <alias>yCol</alias>
                <docs>column Y</docs>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="alternative">
                <value>two.sided</value>
                <docs>type of significance. Possible values are 'two.sided', 'less' or 'greater'</docs>
            </param>
            <param name="mu">
                <value>0</value>
                <docs>assumed means</docs>
            </param>
            <param name="confidenceLevel">
                <value>0.95</value>
                <docs>confidence level</docs>
            </param>
            <param name="paired">
                <docs>True for paired test, otherwise False</docs>
            </param>
            <param name="varEqual">
                <docs>True if variance between two distributions are equal, otherwise False</docs>
            </param>
        </params>
        <ports>
            <port name="x">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>left input data set</docs>
            </port>
            <port name="y">
                <ioType>INPUT</ioType>
                <sequence>2</sequence>
                <type>DATA</type>
                <required>false</required>
                <docs>right input data set. If absent, left input data set will be used.</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="t_test"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_t_test_result"/>
        </metas>
    </algorithm>
    <algorithm codeName="ChiSquare">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        卡方独立性检验是检验两个因素（各有两项或以上的分类）之间是否相互影响的问题，其零假设是两因素之间相互独立。

        %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="xColName">
                <alias>xCol</alias>
                <docs>column X</docs>
            </param>
            <param name="yColName">
                <alias>yCol</alias>
                <docs>column Y</docs>
            </param>
            <param name="probConfig">
                <alias>p</alias>
                <docs>类别概率配置</docs>
            </param>
            <param name="colName">
                <alias>col</alias>
                <docs>需要做卡方检验的列</docs>
            </param>
            <param name="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>summary</outputName>
            </param>
            <param name="outputDetailTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>detail</outputName>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data</docs>
            </port>
            <port name="summary">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
            </port>
            <port name="detail">
                <ioType>OUTPUT</ioType>
                <sequence>2</sequence>
                <type>DATA</type>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="chisq_test"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_chisq_test_result"/>
        </metas>
    </algorithm>
    <algorithm codeName="Covariance">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        用于计算一个矩阵中每一列之间的协方差。

        %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="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="selectedColNames" required="true">
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <alias>cols</alias>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="cov"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_mat_result"/>
        </metas>
    </algorithm>
    <algorithm codeName="MatrixPearson">
        <baseClass>BaseMetricsAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        用于计算一个矩阵中每一列之间的协方差。

        %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="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="selectedColNames" required="true">
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
                <alias>cols</alias>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="corrcoef"/>
            <meta name="xflowProjectName" value="algo_public"/>
            <meta name="calculator" value="$package_root.statistics._customize.get_mat_result"/>
        </metas>
    </algorithm>
    <algorithm codeName="quantile">
        <baseClass>BaseProcessAlgorithm</baseClass>
        <exportFunction>true</exportFunction>
        <docs><![CDATA[
        计算表中几个字段的分位数。

        %params%
        ]]></docs>
        <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="outputTableName">
                <exporter>get_output_table_name</exporter>
                <outputName>output</outputName>
            </param>
            <param name="colName" required="true">
                <alias>cols</alias>
                <exporter>get_feature_columns</exporter>
                <inputName>input</inputName>
            </param>
            <param name="N" required="true">
                <value>100</value>
                <docs>分位点个数，默认 100</docs>
            </param>
        </params>
        <ports>
            <port name="input">
                <ioType>INPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <docs>input data</docs>
            </port>
            <port name="output">
                <ioType>OUTPUT</ioType>
                <sequence>1</sequence>
                <type>DATA</type>
                <schema>
                    <schema>quantile: bigint, {colName}</schema>
                </schema>
            </port>
        </ports>
        <metas>
            <meta name="xflowName" value="quantile"/>
            <meta name="xflowProjectName" value="algo_public"/>
        </metas>
    </algorithm>
</algorithms>