def button_train(name):
scale_img = prepro.collect_data(
os.path.join(os.getcwd(), 'avengers/' + name + '/' + name + '_p.jpg'))
obj = training(modeldir, scale_img, name)
emb_array = obj.main_train()
emb_list = list(map(list, emb_array))
return jsonify({name: emb_list})
def visualization():
for idx, stock in enumerate(config.stock_names):
timestamps = get_timestamps(config.yrs, config.mths, config.dys)
df = collect_data(timestamps, stock, config.moving_averages, True)
fig1 = plot_closing(df, moving_averages=True, intervals=None)
fig1.show()
fig2 = plot_gain(df)
fig2.show()
daily_returns, fig1_c, fig2_c = compare_stocks(config.stock_names_compare, timestamps)
def button_train(name):
img = cv2.imread(
os.path.join(os.getcwd(), 'avengers/' + name + '/' + name + '_p.jpg'))
scale_img = prepro.collect_data(img)
obj = training(modeldir, scale_img, name)
emb_array = obj.main_train()
feature_list.append(feature_map(name, emb_array))
return "Success!"
def get_data(csv):
if csv:
df = pd.read_csv(csv)
df = df.set_index('Date')
return df
else:
timestamps = preprocess.get_timestamps(preprocess.config.yrs,
preprocess.config.mths,
preprocess.config.dys)
df = preprocess.collect_data(
timestamps,
preprocess.config.stock_names[0],
moving_averages=preprocess.config.moving_averages,
include_gain=True)
return df
def make_predictions(features):
timestamps = get_timestamps(config.yrs, config.mths, config.dys)
if len(config.stock_names) == 1:
for feature in features:
df = collect_data(timestamps, config.stock_names[0], moving_averages = config.moving_averages, include_gain=True)
dataset = Dataset(df, feature = feature)
dataset.get_dataset(scale=True)
train_data, test_data, train_data_len = dataset.split(train_split_ratio = 0.8, time_period = 30)
x_train, y_train = train_data
x_test, y_test = test_data
model = keras_lstm(x_train)
pred = Predict()
pred.train([x_train, y_train], model, show_progress=True)
scaler = dataset.scaler
predictions= pred.predict([x_test, y_test], scaler, data_scaled=True, show_predictions=True)
plot_predictions(df, train_data_len, predictions)
def click2():
"""
respond to the button2 click
"""
# toggle button colors as a test
if (button_flag[2] % 2 ==1):
button2.config(bg="white")
button_flag[2] +=1
if button_flag[2] == 1:
#pre_image.append(aug_img)
print ("Training Start")
scale_img = prepro.collect_data(os.path.join(os.getcwd(),'avengers/hermsworth/hermsworth_p.jpg'))
obj = training(modeldir, scale_img, "hermsworth")
get_feature = obj.main_train()
feature_list.append(get_feature)
print('Getting feature map succeed')
elif(button_flag[2] %2 ==0):
button2.config(bg="green")
button_flag[2] += 1