def tree_generate(data, tmp_attributes, attributes, optimizer):
c, same = same_class(data)
node = TreeNode()
# situation 1
if same is True:
node.data = data
node.category = c
return node
if Attribute.empty(tmp_attributes) is True or have_same_attributes(
data, tmp_attributes, attributes):
c = most_class(data)
node.data = data
node.category = c
return node
# attr 可能是continuous的, 此时attr.value有值
attr, weight, data_set = optimizer(data, tmp_attributes, attributes)
# print(attr.name)
node.attribute = attr
node.weight = weight
node.next_node = []
i = 0
if attr.discrete is True:
index = find_attr(attr, attributes)
# print(index)
for iset in data_set:
for d in iset:
if d.value[index] == "?":
d.w *= node.weight[i]
if len(iset) == 0:
new_node = TreeNode(data=iset)
new_node.category = most_class(data)
node.next_node.append(new_node)
else:
Attribute.remove(tmp_attributes, attr)
new_node = tree_generate(iset, tmp_attributes, attributes,
optimizer)
node.next_node.append(new_node)
i += 1
else:
vindex = find_index(attr.range, attr.value)
aindex = find_attr(attr, tmp_attributes)
tmp_attrs1 = copy.deepcopy(tmp_attributes)
tmp_attrs2 = copy.deepcopy(tmp_attributes)
tmp_attrs = [tmp_attrs1, tmp_attrs2]
tmp_attrs1[aindex].range = tmp_attrs1[aindex].range[:vindex]
tmp_attrs2[aindex].range = tmp_attrs2[aindex].range[vindex + 1:]
if len(tmp_attrs1[aindex].range) == 0:
Attribute.remove(tmp_attrs1, attr)
if len(tmp_attrs2[aindex].range) == 0:
Attribute.remove(tmp_attrs2, attr)
for i in range(2):
for d in data_set[i]:
if d.value[aindex] == "?":
d.w *= node.weight[i]
if len(data_set[i]) == 0:
new_node = TreeNode(data=data_set[i])
new_node.category = most_class(data)
node.next_node.append(new_node)
else:
new_node = tree_generate(data_set[i], tmp_attrs[i], attributes,
optimizer)
node.next_node.append(new_node)
return node
