def check_norm_dist(
features,
target,
params,
):
rs = ss(n_splits=150, train_size=.7, random_state=0)
scores_0 = []
scores_1 = []
mean_0 = []
mean_1 = []
S = []
for train_index, _ in rs.split(features):
classifier0 = XGBClassifier(n_estimators=params[0],
max_depth=4,
silent=1)
score0 = cross_val_score(classifier0,
features.iloc[train_index],
target[train_index],
scoring='roc_auc',
cv=10)
scores_0.append(score0)
classifier1 = XGBClassifier(n_estimators=params[1],
max_depth=4,
silent=1)
score1 = cross_val_score(classifier1,
features.iloc[train_index],
target[train_index],
scoring='roc_auc',
cv=10)
scores_1.append(score1)
scores_0 = np.asmatrix(scores_0)
scores_1 = np.asmatrix(scores_1)
for i in range(scores_0.shape[0]):
diff = []
mean_0.append(scores_0[i].mean())
mean_1.append(scores_1[i].mean())
for j in range(scores_0.shape[1]):
diff.append(scores_0[i, j] - scores_1[i, j])
S.append(np.array(diff).var())
means = list((map(lambda x, y: x - y, mean_0, mean_1)))
pylab.subplot(1, 2, 1)
stats.probplot(S, dist="chi2", sparams=(150), plot=pylab)
pylab.subplot(1, 2, 2)
stats.probplot(means, dist="norm", plot=pylab)
pylab.show()
return means, S
def load_data():
X_train = []
X_train_id = []
y_train = []
files_all = []
index = 0
for fl in os.listdir(tr_dir):
fld = os.path.join(tr_dir, fl)
print('load folder {} (Index:{})'.format(fld, index))
path = os.path.join(fld, '*.jpg')
files = sorted(glob.glob(path))
for fl in files:
flbase = os.path.basename(fl)
files_all.append(fl)
X_train_id.append(flbase)
y_train.append(index)
index = index + 1
files_all, y_train = shuffle(files_all, y_train, random_state=9999)
files_all, y_train = shuffle(files_all, y_train, random_state=9999)
print("SHUFFLED TRAINING FILES")
for fl in files_all:
img = get_im_cv2(fl)
X_train.append(img)
print("Collected data TRAINING FILES")
X_train = np.array(X_train, dtype=np.uint8)
y_train = np.array(y_train, dtype=np.uint8)
X_train = X_train.transpose((0, 3, 1, 2))
X_train = X_train.astype('float32')
X_train = X_train / 255
y_flat = y_train
y_train = np_utils.to_categorical(y_train, c)
print("PREPARED TRAINING DATA")
from sklearn.model_selection import StratifiedShuffleSplit as ss
sss = ss(n_splits=1, test_size=0.1, random_state=0)
for train_index, test_index in sss.split(X_train, y_train):
x_t, x_v = X_train[train_index], X_train[test_index]
y_t, y_v = y_train[train_index], y_train[test_index]
y_f, y_f_v = y_flat[train_index], y_flat[test_index]
return x_t, y_t, x_v, y_v, y_f
def __init__(self,
Cs=500,
cv=10,
sampler='skf',
solver='liblinear',
**kwargs):
super(self.__class__, self).__init__()
self.penalty = 'l1'
self.solver = solver
self.Cs = Cs
self.sampler = sampler
self.cv_folds = cv
if self.sampler == 'skf':
self.cv = skf(n_splits=self.cv_folds)
elif self.sampler == 'sss':
self.cv = sss(n_splits=self.cv_folds)
elif self.sampler == 'kf':
self.cv = kf(n_splits=self.cv_folds)
elif self.sampler == 'ss':
self.cv = ss(n_splits=self.cv_folds)
else:
raise (Exception(
'Selected sampler is not a valid. Please choose '
'"skf" for stratified K-fold or "sss" for '
'stratified shuffle split. Also "sk" and "ss" for '
'the respective non-stratified methods.'))
for k, v in kwargs.items():
setattr(self, k, v)
self.x = None
self.y = None
def cv_reg(x, test_size = 0.2, n_splits = 5, random_state=None):
return ss(n_splits, test_size, random_state=random_state).split(x)
def cv_reg(x, test_size = 0.2, n_splits = 5, random_state=None): return ss(n_splits, test_size, random_state=random_state).split(x) def timeit(klass, params, x, y):