def test_progressbar(max_val, sleep_time=2):
bar = Scoring.Bar(max_val)
bar.start()
for i in range(max_val):
bar.update(i + 1)
sleep(sleep_time)
bar.finish()
print("all is good")
def GridSearchPCA(estimator,
params,
Data,
config,
n_comp,
name,
location,
states=3):
df = CreateDataFrame(Data=Data, config=config)
KFold = FE.KFold(Data.shape[2])
start = time()
GRID = ParameterGrid(params)
combinations = len(list(GRID))
print("Number of combinations {}".format(combinations))
bar = Scoring.Bar(combinations * 10)
combo = 0
bar.start()
for i, g in enumerate(GRID):
score_tab = Scoring.ScoringTable(location=location,
name=name + str(g) + str(i),
n_states=states)
for cross in range(Data.shape[0]):
clf = copy(estimator)
X_train, y_train, X_test, y_test = KFold.fit_transform(
x=df, kFoldIndex=cross)
scale = StandardScaler()
X_train = scale.fit_transform(X=X_train, y=y_train)
X_test = scale.transform(X=X_test)
pca = PCA(n_components=n_comp)
X_train = pca.fit_transform(X=X_train)
X_test = pca.transform(X=X_test)
clf.set_params(**g)
pred = train_and_predict(model=clf,
train=X_train,
test=X_test,
labels=y_train,
unsupervised=False)
score = Scoring.score(states=pred,
results=np.array(y_test),
unsupervised=False,
pocet_stavu=states)
score_tab.add(scores=score)
combo += 1
bar.update(combo)
score_tab.save_table()
del score_tab
info = copy(config)
info["params"] = g
with open(location + name + str(g) + str(i) + '_config.pickle',
'wb') as f:
pickle.dump(info, f)
bar.finish()
print('Celý proces trval: {} vteřin'.format(
np.around(time() - start, decimals=0)))
print('Hotovo!!')
return