import data_sort
from sklearn.ensemble import AdaBoostClassifier
boost = AdaBoostClassifier(algorithm = "SAMME", n_estimators = 100)
number_of_features = 15
training_data = data_sort.makeSet("../datasets/adult_data_big.txt", number_of_features)
test_data = data_sort.makeSet("../datasets/adult_data_test.txt", number_of_features)
X_train, y_train = data_sort.split(training_data)
X_test, y_test = data_sort.split(test_data)
boost.fit(X_train, y_train)
print(boost.score(X_test, y_test))
# adaptive_boosting.py modifed for (somewhat) more efficient testing
import data_sort
from dt_thread import TreeThread
from random import randint
import math
import time
number_of_workers_list = [200, 400]
number_of_features = 7
training_data = data_sort.makeSet("datasets/car_split_train.txt", number_of_features)
training_data, labels = data_sort.binaryfy(training_data)
print(len(training_data))
test_data = data_sort.makeSet("datasets/car_split_test.txt", number_of_features)
test_data, labels = data_sort.binaryfy(test_data)
# Loss function only indirectly used
# def loss_function(test_data):
# loss = 0
# for n in range(0, len(test_data)):
# m = 0
# for k in range(0, len(threads)):
# m += alpha[k]*threads[k].binary_query(test_data[n])
# m = m*test_data[n][-1]
# loss += exp(-m)
# return loss
def calculate_weights(weights, error_rate, alpha, data_set, predictions):
temp_weights = weights.copy()
import data_sort
training_data = data_sort.makeSet("adult_data_big.txt", 15)
label1 = 0
label2 = 0
file = open("adult_data_unbiased.txt", "w+")
for a in training_data:
if label1 != 0:
if a[14] != label1[14]:
label2 = a
file.write(str(label1) + "\n")
file.write(str(label2) + "\n")
label1 = 0
label2 = 0
else:
label1 = a
file.close()
startTime = time.time()
training_data = [['Green', 3, 'Apple'], ['Red', 3,
'Apple'], ['Red', 1, 'Grape'],
['Red', 2, 'Grape'], ['Yellow', 3, 'Lemon'],
['Yellow', 3, 'Apple'], ['Red', 1, 'Grape'],
['Red', 2, 'Grape'], ['Green', 1, 'Grape'],
['Yellow', 2, 'Lemon']]
test_data = [['Green', 1, 'Grape'], ['Yellow', 2, 'Lemon'],
['Green', 2, 'Apple'], ['Green', 4, 'Apple']]
number_of_workers = 10
data_for_workers = []
number_of_features = 15
training_data = data_sort.makeSet("datasets/adult_data.txt",
number_of_features)
weights = []
for i in range(0, len(training_data)):
weights.append(1)
# TODO: Ta hänsyn till weights, görs i adaptive_boosting.py
def extract_training_data(dataSet):
data = []
for n in range(0, int(round(0.6 * len(dataSet)))):
data.append(dataSet[randint(0, len(dataSet) - 1)])
return data
# Create threads