Question

在我的项目中，我必须在交叉验证中使用SVM classifer和Pipeline进行特征选择。特征选择方法是基于相关的特征选择（我通过python-weka-wrapper使用Weka库）。所以，我用 fit（）和 Transform（）方法编写了这个类：

import numpy as np
from sklearn.base import BaseEstimator, TransformerMixin
from convertArff import  arffOutput
import weka.core.jvm as jvm

from weka.attribute_selection import ASSearch, ASEvaluation, AttributeSelection
from weka.core.converters import Loader

class CorrelationFeatureSelection(BaseEstimator, TransformerMixin):


    def __init__(self, names, array):
        self.names = names
        self.array = array



    def _reset(self):
        """Reset internal data-dependent state of the scaler, if       necessary.
        __init__ parameters are not touched.
        """

        #  Checking one attribute is enough, becase they are all set together
        # in partial_fit
        if hasattr(self, 'attibutes_selected_'):
            del self.attributes_selected_




    def fit(self, X, y=None):
        self._reset()
        print type(self.array)
        arffOutput("result", self.array, self.names)
        jvm.start()
        print "sono dentro"
        loader = Loader(classname="weka.core.converters.ArffLoader")
        data = loader.load_file("result.arff")
        data.class_is_last()

        print(data)

        search = ASSearch(classname="weka.attributeSelection.BestFirst", options=["-D", "1", "-N", "5"])
        evaluator = ASEvaluation(classname="weka.attributeSelection.CfsSubsetEval", options=["-P", "1", "-E", "1"])
        attsel = AttributeSelection()
        attsel.search(search)
        attsel.evaluator(evaluator)
        attsel.select_attributes(data)

        print("# attributes: " + str(attsel.number_attributes_selected))
        print("attributes: " + str(attsel.selected_attributes))
        print("result string:\n" + attsel.results_string)
        attributes = attsel.selected_attributes
        jvm.stop()

        print attributes
        print type(attributes)
        self.attributes_selected_ = attributes[0:len(attributes) - 1]
        print self.attributes_selected_
        return self

    def transform(self, X):

        col_list = []
        for c in self.attributes_selected_:
            col_list.append(X[:, c:c + 1])
        return np.concatenate(col_list, axis=1)

当我运行我的代码时，我收到以下错误：

RuntimeError: Failed to start Java VM

这个问题有解决办法吗？

Answer 1

启动和停止JVM应该在启动应用程序的main方法中发生（即在if __name__ == "__main__":块中），因为不幸的是，你不能多次启动/停止JVM。

以下代码将在第二个jvm.start()失败：

import weka.core.jvm as jvm

print("Starting 1")
jvm.start()
print("Stopping 1")
jvm.stop()
print("Starting 2")
jvm.start()
print("Stopping 2")
jvm.stop()

管道内的Python-Java虚拟机

1 个答案: