def test_DTA_membership(self):
s = "and(or(and(true()false())true())and(true()or(false()true())))"
tree = Tree.parse(s)
tree.write_graphviz("test.txt")
self.assertEqual(self.DTA.membership(tree), Result.accept)
self.assertEqual(self.DTA.membership(Tree.parse("not(false())")),
Result.accept)
def cut_discriminator(self, trees, value):
if value < 0 or value > 100:
Tools.print_log_line('Value must debe ser un valor entre 0 y 100',
logging.ERROR)
else:
return Tree.sum_tree_list(trees, 'expan') * ((100 - value) / 100.0)
return 0
def test_duta_membership(self):
s = "and(or(and(true()false())true())and(true()or(false()true()true()false())))"
tree = Tree.parse(s)
# print tree
# print self.drta.membership(tree)
self.assertEqual(self.duta.membership(tree), Result.accept)
def apply_model(self, plot: Plot, years: int, value: float):
Tools.print_log_line('Aplicando corta por el menor', logging.INFO)
accumulator = 0
cut_all_the_rest = False
new_plot = Plot()
new_plot.clone(plot)
order_criteria = OrderCriteria(ASC)
order_criteria.add_criteria('dbh')
search_criteria = SearchCriteria()
search_criteria.add_criteria('status', None, EQUAL)
trees = Tree.get_sord_and_order_tree_list(plot.trees,
search_criteria=search_criteria,
order_criteria=order_criteria)
cut_discriminator = self.type.cut_discriminator(trees, value)
for tree in trees:
accumulator += self.type.accumulator(tree)
if not cut_all_the_rest:
new_tree = Tree()
new_tree.clone(tree)
if accumulator >= cut_discriminator:
cut_all_the_rest = True
new_expan = self.type.compute_expan(tree, accumulator, cut_discriminator)
if new_expan <= 0:
new_tree.add_value('expan', new_expan)
new_tree.add_value('status', 'C')
else:
cut_tree = Tree()
cut_tree.clone(tree)
cut_tree.add_value('status', 'C')
cut_tree.add_value('expan', new_expan)
if cut_tree.expan > 0:
new_plot.add_tree(cut_tree)
new_tree.sub_value('expan', new_expan)
new_tree.add_value('status', None)
new_plot.add_tree(new_tree)
else:
new_tree = Tree()
new_tree.clone(tree)
new_tree.add_value('status', 'C')
new_plot.add_tree(new_tree)
return new_plot
def generate_y_sketch(y):
y = str(y)
if not y:
return y
return Tree(y).get_compact_form(True)
def generate_y_compact(y):
y = str(y)
if not y:
return y
return Tree(y).get_compact_form()
def apply_model(self, plot: Plot, years: int, value: float):
Tools.print_log_line('Aplicando cut down sistemática', logging.INFO)
new_plot = Plot()
new_plot.clone(plot)
order_criteria = OrderCriteria()
order_criteria.add_criteria('dbh')
search_criteria = SearchCriteria()
search_criteria.add_criteria('status', None, EQUAL)
trees = Tree.get_sord_and_order_tree_list(plot.trees,
search_criteria=search_criteria,
order_criteria=order_criteria)
for tree in trees:
new_tree = Tree()
new_tree.clone(tree)
new_tree.add_value('expan', tree.expan * ((100 - value) / 100))
new_tree.add_value('status', None)
new_plot.add_tree(new_tree)
cut_tree = Tree()
cut_tree.clone(tree)
cut_tree.add_value('status', 'C')
cut_tree.add_value('expan', tree.expan - new_tree.expan)
if cut_tree.expan > 0:
new_plot.add_tree(cut_tree)
return new_plot
def evaluate_predictions(gold_filename: str, mode: str) -> Dict:
instance_count: int = 0
exact_matches: int = 0
non_terminal_matches: int = 0
terminal_matches: int = 0
invalid_preds: float = 0
labeled_bracketing_scores = Calculator(strict=False)
tree_labeled_bracketing_scores = Calculator(strict=True)
gold_list = []
pred_list = []
exact_match_list = []
tree_depth_list = []
tree_size_list = []
with open(gold_filename) as gold_file:
# for gold_line, pred_line in zip_longest(gold_file, pred_file):
for gold_line in gold_file:
try:
gold_lines = gold_line.strip().split("\t")
gold_line = gold_lines[1]
pred_line = gold_lines[2] #pred_line.strip()
except AttributeError:
print("WARNING: check format and length of files")
quit()
(non_terminal_match,
terminal_match) = comp_gold_and_pred(gold_line, pred_line)
non_terminal_matches += non_terminal_match
terminal_matches += terminal_match
try:
gold_tree = Tree(gold_line)
instance_count += 1
except ValueError:
print("FATAL: found invalid line in gold file:", gold_line)
quit()
try:
pred_tree = Tree(pred_line)
labeled_bracketing_scores.add_instance(gold_tree, pred_tree)
tree_labeled_bracketing_scores.add_instance(
gold_tree, pred_tree)
except ValueError:
# print("WARNING: found invalid line in pred file:", pred_line)
invalid_preds += 1
# (non_terminal_match, terminal_match) = comp_gold_and_pred(gold_line, pred_line)
# non_terminal_matches += non_terminal_match
# terminal_matches += terminal_match
labeled_bracketing_scores.add_instance(gold_tree)
tree_labeled_bracketing_scores.add_instance(gold_tree)
continue
if str(gold_tree) == str(pred_tree):
exact_matches += 1
exact_match_list.append(1)
else:
exact_match_list.append(0)
if mode == 'sketch':
tree_depth_list.append(gold_tree.root.get_depth() - 1)
else:
tree_depth_list.append(gold_tree.root.get_depth() - 2)
tree_size_list.append(gold_tree.root.get_size())
gold_list.append(gold_line)
pred_list.append(pred_line)
# if gold_tree.get_compact_form(sketch=True) == pred_tree.get_compact_form(sketch=True):
# non_terminal_matches += 1
#
# if gold_tree.root.get_str_token_spans() == pred_tree.root.get_str_token_spans():
# terminal_matches += 1
result_file = mode + '.final_result.tsv'
result = {
'gold': gold_list,
'pred': pred_line,
'match': exact_match_list,
'size': tree_size_list,
'depth': tree_depth_list
}
result_df = pd.DataFrame.from_dict(result)
result_df.to_csv(result_file, sep='\t', header=True, index=False)
exact_match_fraction: float = (exact_matches /
instance_count) if instance_count else 0
tree_validity_fraction: float = (
1 - (invalid_preds / instance_count)) if instance_count else 0
# print("terminal list: {}".format(terminal_list))
return {
"instance_count": instance_count,
"invalid_preds": invalid_preds,
"exact_match_count": exact_matches,
"exact_match_accu": exact_match_fraction,
"non_terminal_match_Error_count":
(instance_count - non_terminal_matches),
"terminal_match_Error_count": (instance_count - terminal_matches),
"total_Error_count": (instance_count - exact_matches)
# "labeled_bracketing_scores":
# labeled_bracketing_scores.get_metrics(),
# "tree_labeled_bracketing_scores":
# tree_labeled_bracketing_scores.get_metrics(),
# "tree_validity":
# tree_validity_fraction
}