Beispiel #1
0
def test_top_n_counts():
    '''Test the top n counts method'''
    ngram_freq = NgramFrequencies(1, '')
    ngram_freq.dic_ngram["ab"] = 3
    ngram_freq.dic_ngram["cd"] = 2
    ngram_freq.dic_ngram["ef"] = 6
    list_top_count = ngram_freq.top_n_counts(3)
    assert list_top_count[0][0] == "ef"
    assert list_top_count[1][0] == "ab"
    assert list_top_count[2][0] == "cd"
Beispiel #2
0
def test_ngram_frequencies():
    nf = NgramFrequencies(2)
    assert nf.add_item("He_is") == 1
    assert nf.add_item("He_is") == 2
    assert nf.add_item("He_is") == 3
    assert nf.add_item("I_am") == 1
    assert nf.add_item("I_am") == 2
    assert nf.add_item("I_am") == 3
    assert nf.add_item("I_am") == 4
    assert nf.add_item("I_am") == 5
    assert nf.add_item("I_am") == 6
    assert nf.add_item("They_are") == 1

    assert nf.frequency("They_are") == 0.1

    assert nf.top_n_counts() == [("I_am", 6), ("He_is", 3)]
    assert nf.top_n_freqs() == [("I_am", 0.6), ("He_is", 0.3)]
Beispiel #3
0
def test_top_n_counts():
    """test returning a list of items sorted on the count"""
    ngrams = NgramFrequencies()
    new_dic = {"a": 1, "b": 2, "c": 3, "d": 4}
    top_list = ngrams.top_n_counts(new_dic)
    assert top_list == [("d", 4), ("c", 3), ("b", 2), ("a", 1)]