def make_votes(datasets, binary, clf):
if (type(clf) == list):
votes = []
for clf_i in clf:
votes += [
learn.train_model(data_i, binary, clf_i) for data_i in datasets
]
else:
votes = [learn.train_model(data_i, binary, clf) for data_i in datasets]
if (binary):
votes = [one_hot_result(vote_i) for vote_i in votes]
return votes
def visualize_corl(paths):
datasets = ens.read_dataset(paths["common"], paths["binary"])
results = [learn.train_model(data_i) for data_i in datasets]
d = [result_i.true_one_hot() for result_i in results]
C = corl(results, d)
C = np.round_(C, decimals=3)
heat_map(C)
def validate_acc(data_i):
train=data_i.split()[0]
rename_i={name_i:"%d_%d_%d" % (name_i.get_cat()+1,i%2,i)
for i,name_i in enumerate(train.keys())}
new_data=train.rename(rename_i)
result=learn.train_model(new_data,binary=False,clf_type="LR")
return result#.get_acc()
def cross_acc(datasets):
datasets=[data_i.split()[0] for data_i in datasets]
for data_i in datasets:
data_i.info=["%s_%d" %(name_i.split('_')[0],i)
for i,name_i in enumerate(data_i.info)]
acc=[ learn.train_model(data_i,
binary=True,clf_type="LR",acc_only=True)
for data_i in datasets]
return acc
def make_votes(common_path, binary_path, clf="LR", read=None):
if (read is None):
read = read_dataset
datasets = read(common_path, binary_path)
if (len(datasets) == 0):
raise Exception("No data at %s" % binary_path)
results = [
learn.train_model(data_i, clf_type=clf, binary=False)
for data_i in datasets
]
return Votes(results)
def ensemble(paths, system=None, clf="LR", s_clf=None, transform=None):
datasets = ens.read_dataset(paths["common"], paths["binary"])
if (transform):
datasets = [transform(data_i) for data_i in datasets]
results = [
learn.train_model(data_i, clf_type=clf, binary=False)
for data_i in datasets
]
if (s_clf):
results = [results[clf_i] for clf_i in s_clf]
votes = ens.Votes(results)
return voting(votes, system)
def __call__(self, datasets, clf="LR"):
results = []
new_datasets = []
for data_i in datasets:
train_i, test_i = data_i.split()
s_train_i = self.subsample(train_i)
s_data_i = {**s_train_i, **test_i}
s_data_i = feats.Feats(s_data_i)
new_datasets.append(s_data_i)
result_i = learn.train_model(s_data_i, binary=False, clf_type=clf)
results.append(result_i)
return new_datasets, results
def ens_acc(paths, clf="LR", acc_only=True):
if (type(paths) == tuple):
datasets = tools.read_datasets(paths)
return [
learn.train_model(data_i, False, clf, acc_only)
for data_i in datasets
]
votes = learn.read_votes(paths)
y_true = [int(name_i.split("_")[0]) - 1 for name_i in votes[0][2]]
result = [learn.voting([vote_i], False) for vote_i in votes]
acc = [accuracy_score(y_true, result_i) for result_i in result]
return acc
def __call__(self, paths, binary=False, clf="LR", s_clf=None):
datasets = self.read(paths["common"], paths["binary"])
if (self.transform):
datasets = [self.transform(data_i) for data_i in datasets]
results = [
learn.train_model(data_i, clf_type=clf, binary=False)
for data_i in datasets
]
votes = Votes(results)
if (s_clf):
votes = Votes([votes.results[i] for i in s_clf])
result = votes.voting(binary)
print(result.get_acc())
return result, votes
def person_acc(data_i):
train = data_i.split()[0]
persons = set([name_i.get_person() for name_i in train.keys()])
results = []
print("person acc")
for j in list(persons):
def helper(name_i):
cat_i = name_i.get_cat() + 1
person_i = int(name_i.get_person() != j)
return "%d_%d" % (cat_i, person_i)
rename_j = {
name_i: "%s_%d" % (helper(name_i), i)
for i, name_i in enumerate(train.keys())
}
train_j = train.rename(rename_j)
result_j = learn.train_model(train_j, binary=False, clf_type="LR")
results.append(result_j)
return unify_results(results)
import data_generator
import learn
if __name__ == '__main__':
# Generate sample data
data_generator.generate_data()
# Train the linear regression model
learn.train_model()
