在Python3.6中调用sklearn2pmml()函数会抛出RuntimeError

Question

我正在尝试将管道对象保存为PMML，而Python抛出了一个RuntimeError。

我的Python版本是3.6，sklearn2pmml版本是0.44.0，JDK版本是1.8.0_201。

所有这些都符合软件包的前提条件。

以下是我到目前为止所做的工作。(我不包括数据加载和清理部分)

from sklearn2pmml.pipeline import PMMLPipeline
from sklearn2pmml import make_pmml_pipeline, sklearn2pmml

logit_pipline = Pipeline([('vect', CountVectorizer(ngram_range=(1,2))), ('tfidf', TfidfTransformer(use_idf=True)), ('clf', LogisticRegression(C=11.3))])
pmml_pipeline = PMMLPipeline([("logit", logit_pipline)])
pmml_pipeline.fit(X, Y)

sklearn2pmml(pmml_pipeline, 'logit.pmml', with_repr=True)

在我运行上面提到的最后一行之后发生的事情是...

sklearn2pmml(pmml_pipeline, 'logit.pmml', with_repr=True)
Standard output is empty
Standard error:
Apr 30, 2019 11:59:04 AM org.jpmml.sklearn.Main run
INFO: Parsing PKL..
Apr 30, 2019 11:59:04 AM org.jpmml.sklearn.Main run
INFO: Parsed PKL in 230 ms.
Apr 30, 2019 11:59:04 AM org.jpmml.sklearn.Main run
INFO: Converting..
Apr 30, 2019 11:59:04 AM org.jpmml.sklearn.Main run
SEVERE: Failed to convert
java.lang.IllegalArgumentException: Expected an estimator object as the last step, got a transformer object (Python class sklearn.pipeline.Pipeline)
        at sklearn2pmml.pipeline.PMMLPipeline.getEstimator(PMMLPipeline.java:541)
        at sklearn2pmml.pipeline.PMMLPipeline.encodePMML(PMMLPipeline.java:93)
        at org.jpmml.sklearn.Main.run(Main.java:145)
        at org.jpmml.sklearn.Main.main(Main.java:94)

Exception in thread "main" java.lang.IllegalArgumentException: Expected an estimator object as the last step, got a transformer object (Python class sklearn.pipeline.Pipeline)
        at sklearn2pmml.pipeline.PMMLPipeline.getEstimator(PMMLPipeline.java:541)
        at sklearn2pmml.pipeline.PMMLPipeline.encodePMML(PMMLPipeline.java:93)
        at org.jpmml.sklearn.Main.run(Main.java:145)
        at org.jpmml.sklearn.Main.main(Main.java:94)

Traceback (most recent call last):

  File "<ipython-input-129-f5c307b4aaba>", line 1, in <module>
    sklearn2pmml(pmml_pipeline, 'logit.pmml', with_repr=True)

  File "C:\ProgramData\Anaconda3\lib\site-packages\sklearn2pmml\__init__.py", line 252, in sklearn2pmml
    raise RuntimeError("The JPMML-SkLearn conversion application has failed. The Java executable should have printed more information about the failure into its standard output and/or standard error streams")

RuntimeError: The JPMML-SkLearn conversion application has failed. The Java executable should have printed more information about the failure into its standard output and/or standard error streams

现在根据一些人的说法，这是一些JDK兼容性问题，使用JDK版本1.9和更高版本或1.6和更低版本会引发这种问题。但是既然我的JDK版本是sklearn2pmml可以接受的，为什么会出现这种错误呢？

Answer 1

正如底层Java异常所表明的那样，sklearn2pmml.pipeline.PMMLPipeline类需要使用步骤列表进行参数化，其中最后一步包含一些估计器对象。在本例中，您使用一个单元素的步骤列表对PMMLPipeline进行参数化；最后一步包含一个Pipeline对象，从这个意义上讲，它不是一个估计器对象。

要解决这个问题，只需去掉中间的logit_pipline层(将管道包装在管道中到底是什么意思呢？)。

例如，这是可行的：

logit_pipline = PMMLPipeline([..])
logit_pipeline.fit(X, y)
sklearn2pmml(logit_pipeline, "logit.pmml")

这个问题与JDK、Python或Scikit-Learn版本完全无关。