def test_viterbi(self):
x_thin = pd.Series(['*', '*', 'The', 'Treasury', 'is', '<STOP>'])
y_thin = pd.Series(['*', '*', 'DT', 'NNP', 'VBZ', '<STOP>'])
tests = [f'f_10{x_}'
for x_ in range(8)] + [f'tri_00{x_}' for x_ in range(7)]
data = PreprocessTags().load_data(r'..\data\test.wtag')
model1 = Model(tests)
model1.x = x
model1.y = y
base_corpus = pd.Series(['*', '<STOP>'])
tag_corpus = pd.Series(y.value_counts().drop(['*', '<STOP>']).index)
model1.tag_corpus = base_corpus.append(tag_corpus)
model1.tag_corpus_tokenized = range(len(model1.tag_corpus))
model1._translation() # create dictionaries for tokenizing
model1._vectorize()
#
# below result of fit on full data (model1.fit(data.x, data.y))
#
model1.v = [
9.98440989e-04, 4.55460621e+00, 3.50976884e+00, 9.98440989e-04,
9.98440989e-04, 9.98440989e-04, 9.98440989e-04, 9.98440989e-04,
3.53744043e+00, 4.72940057e+00, 2.86124632e+00, 3.02403509e+00,
9.98440989e-04, 2.75377462e+00, 9.98440989e-04
]
# print(model1.vector_x_y)
# print(model1.lin_loss_matrix_x_y)
b = model1.predict(x)
# b = model1._viterbi(x)
print("viterbi result")
print(b)
