def check_boosting():
np.random.seed(5)
test_classifiers = set()
s_set = {1, -1}
b_set = {1, 3, 5, 7, 9, 11}
for s in s_set:
for b in b_set:
test_classifiers.add(DecisionStump(s, b, 0))
test_features = [[2], [4], [6], [8], [10]]
test_labels = [1, 1, -1, -1, 1]
try:
import boosting
except:
return 0
try:
test_ada1 = boosting.AdaBoost(test_classifiers, 1)
test_ada2 = boosting.AdaBoost(test_classifiers, 3)
test_ada1.train(test_features, test_labels)
test_ada2.train(test_features, test_labels)
predictions_ada1 = test_ada1.predict(test_features)
predictions_ada2 = test_ada2.predict(test_features)
if predictions_ada1 == [1, 1, -1, -1, -1
] and predictions_ada2 == [1, 1, -1, -1, 1]:
score_ada = 0.5
else:
score_ada = 0
except:
score_ada = 0
try:
test_logit1 = boosting.LogitBoost(test_classifiers, 1)
test_logit2 = boosting.LogitBoost(test_classifiers, 3)
test_logit1.train(test_features, test_labels)
test_logit2.train(test_features, test_labels)
predictions_logit1 = test_logit1.predict(test_features)
predictions_logit2 = test_logit2.predict(test_features)
if predictions_logit1 == [1, 1, -1, -1, -1] and predictions_logit2 == [
1, 1, -1, -1, 1
]:
score_logit = 0.5
else:
score_logit = 0
except:
score_logit = 0
return round(score_ada + score_logit, 1)
# print(parts[:i]+parts[i+1:])
d, lb = list(
map(np.array,
zip(*(sum(parts[:i] + parts[i + 1:], start=[])))))
self.classifiers[i].fit(d, lb)
# print(parts[i])
acc = np.mean([
int(self.classifiers[i].predict(data_point) == label)
for data_point, label in parts[i]
])
accuracies.append(acc)
return np.mean(accuracies)
if __name__ == '__main__':
iris = datasets.load_iris()
vectorized_map = np.vectorize(lambda x, mean: x > mean)
tmp = iris['data']
data = np.array(
[vectorized_map(tmp[:, i], np.mean(tmp[:, i])) for i in range(4)]).T
labels = iris['target']
data, labels = list(zip(*shuffle(list(zip(data[:99], labels[:99])))))
data = np.array(data)
labels = np.array(labels)
classifier = boosting.AdaBoost(DecisionTreeClassifier, 5, .5, sklearn=True)
labels = classifier.change_labels(labels, possible_labels={1, 0, -1})
validator = CrossValidation(5, classifier)
print(validator.evaluate(data, labels))
assert(len(y_test) == 20)
# set classifiers
h_set = set()
s_set = {1, -1}
b_set = set(np.linspace(0, 10, 51))
d_set = {0, 1, 2, 3}
for s in s_set:
for b in b_set:
for d in d_set:
h_set.add(decision_stump.DecisionStump(s,b,d))
# training
Adas = []
for idx, T in enumerate([10, 20, 30]):
Adas.append(boosting.AdaBoost(h_set, T=T))
Adas[idx].train(X_train, y_train)
# testing
Ada_preds = []
Ada_accus = []
for Ada in Adas:
pred = Ada.predict(X_test)
Ada_preds.append(pred)
Ada_accus.append(accuracy_score(pred, y_test))
print('AdaBoost testing accuracies:', Ada_accus)
# save
json.dump({'Ada_preds': Ada_preds, 'Ada_accus': Ada_accus},
d in {0, 1}
'''
# data
features_1, labels_1 = data_loader.toy_data_1()
# clfs
h_set = set()
s_set = {1, -1}
b_set = {-2, -0.5, 0.5, 2}
d_set = {0, 1}
for s in s_set:
for b in b_set:
for d in d_set:
h_set.add(decision_stump.DecisionStump(s, b, d))
# boost
Ada_1 = boosting.AdaBoost(h_set, T=1)
Ada_1.train(features_1, labels_1)
# check
print('━━━━━━━━━━ Toy example 1 ━━━━━━━━━━')
print('This toy example checks the format. Any of the stump is correct.')
print('(Can you explain why?)')
print('Ada_1: s = {:01d}, b = {:.1f}, d = {:01d}'.format(
Ada_1.clfs_picked[0].s, Ada_1.clfs_picked[0].b, Ada_1.clfs_picked[0].d))
if Ada_1.betas[0] == 0:
print('Betas are correct')
else:
print('▁▂▃▄▅▆▇█ Betas are not correct █▇▆▅▄▃▂▁')
#################
