Python make_pmi_matrix示例

示例#1

文件： test_pmi.py 项目： Sandy4321/document_feature_selection

def test_make_csr_main():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(input_dict, logger)
    assert isinstance(pmi_document_freq_csr_matrix, csr_matrix)
    assert isinstance(label_group_dict, dict)
    assert isinstance(vocabulary, dict)


    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(input_dict, logger, ngram=2)
    assert isinstance(pmi_document_freq_csr_matrix, csr_matrix)
    assert isinstance(label_group_dict, dict)
    assert isinstance(vocabulary, dict)

示例#2

def test_make_csr_main():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(
        input_dict, logger)
    assert isinstance(pmi_document_freq_csr_matrix, csr_matrix)
    assert isinstance(label_group_dict, dict)
    assert isinstance(vocabulary, dict)

    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(
        input_dict, logger, ngram=2)
    assert isinstance(pmi_document_freq_csr_matrix, csr_matrix)
    assert isinstance(label_group_dict, dict)
    assert isinstance(vocabulary, dict)

示例#3

def test_fit_transform_pmi():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(
        input_dict, logger, ngram=2)
    pmi_featured_csr_matrix = fit_format(pmi_document_freq_csr_matrix,
                                         vocabulary, label_group_dict)
    assert isinstance(pmi_featured_csr_matrix, csr_matrix)
    print(pmi_featured_csr_matrix.toarray())

示例#4

文件： test_pmi.py 项目： Sandy4321/document_feature_selection

def test_pmi_calc():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(input_dict, logger, ngram=2)
    pmi_score_objects = pmi_single_process_main(pmi_document_freq_csr_matrix, vocabulary, label_group_dict, logger, cut_zero=True)
    assert isinstance(pmi_score_objects, list)
    assert isinstance(pmi_score_objects[0], dict)
    assert pmi_score_objects[0].has_key('score')
    assert pmi_score_objects[0].has_key('word')
    assert pmi_score_objects[0].has_key('label')
    print(pmi_score_objects)

示例#5

def test_pmi_calc():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(
        input_dict, logger, ngram=2)
    pmi_score_objects = pmi_single_process_main(pmi_document_freq_csr_matrix,
                                                vocabulary,
                                                label_group_dict,
                                                logger,
                                                cut_zero=True)
    assert isinstance(pmi_score_objects, list)
    assert isinstance(pmi_score_objects[0], dict)
    assert pmi_score_objects[0].has_key('score')
    assert pmi_score_objects[0].has_key('word')
    assert pmi_score_objects[0].has_key('label')
    print(pmi_score_objects)

示例#6

文件： test_pmi.py 项目： Sandy4321/document_feature_selection

def test_fit_transform_pmi():
    pmi_document_freq_csr_matrix, label_group_dict, vocabulary = make_pmi_matrix(input_dict, logger, ngram=2)
    pmi_featured_csr_matrix = fit_format(pmi_document_freq_csr_matrix, vocabulary, label_group_dict)
    assert isinstance(pmi_featured_csr_matrix, csr_matrix)
    print(pmi_featured_csr_matrix.toarray())