I have a question about creating LSTM Memory cell? About node size setting~!!!

x_train = raw_train.loc[:, train_cols].values
y_train = raw_train.loc[:, test_cols].values

x_test = raw_test.loc[:, train_cols].values
y_test = raw_test.loc[:, test_cols].values

min_max_scaler = MinMaxScaler()
min_max_scaler_label = MinMaxScaler()

x_train_scaled = min_max_scaler.fit_transform(x_train)
y_train_scaled = min_max_scaler_label.fit_transform(y_train)

x_test_scaled = min_max_scaler.transform(x_test)
y_test_scaled = min_max_scaler_label.transform(y_test)

def build_timeseries(mat, time_steps):
    dim_0 = mat.shape[0] - time_steps
    dim_1 = mat.shape[1]

    x = np.zeros((dim_0, time_steps, dim_1))
    y = np.zeros((dim_0,))

    for i in range(dim_0):
        x[i] = mat[i:time_steps + i]
        y[i] = mat[time_steps + i, 0]
    print('length of time-series i/o', x.shape, y.shape)
    return x,y

time_step = 1

x_train_scaled_time, y_train_scaled_time = build_timeseries(x_train_scaled, 1)
x_test_scaled_time, y_test_scaled_time = build_timeseries(x_test_scaled, 1)

def create_model(input_data):
    lstm_model = Sequential()
    lstm_model.add(LSTM(10, input_shape=(x_train_scaled_time.shape[1],x_train_scaled_time.shape[2])))
    lstm_model.add(Dense(1))
    lstm_model.compile(loss='mse', optimizer='adam')

    return lstm_model