def standardize_all_data():
# Standardize all data
# Let data's mean = 0
# standardard deviation = 1
# If data are not loaded, this function will load data first
# Once all data are standardized,
# function get_trajectory will return standardized data
if config.STANDARDIZED is True:
return get_trajectory()
if config.ALL_TRAJECTORY is None:
# Load data set
get_trajectory()
data = config.ALL_TRAJECTORY
mean = data_mean()
std = data_std()
for i in range(len(data)):
data[i] = list(data[i])
# First two elements are matadata, 3rd is data
data[i][3] = (data[i][3] - mean) / std
data[i] = tuple(data[i])
config.ALL_TRAJECTORY = data
config.STANDARDIZED = True
return get_trajectory()
def sample_net(net, n):
samples = []
for i in range(n):
sample = get_real_example(net)
sample, _, _ = standardize_data(sample,
mean=data_mean(),
std=data_std())
samples.append(sample)
return samples
def draw_spectra_of(path):
x = _read_one_file(path)
x, _, _ = standardize_data(x, data_mean(), data_std())
x = fft_data([x])[0]
x = np.absolute(x)
plt.plot(x)
plt.show()
return x
def istandardize_data(data):
# Recover standardized data
# The parameter 'data' should be a list of real numpy array
# element in data has shape length * dimensionality
if config.STANDARDIZED is False:
raise Exception('Data set has not been standardized!')
mean = data_mean()
std = data_std()
recovered = []
for d in data:
r = d * std + mean
recovered.append(r)
return recovered
a = np.array([[1, 2], [1, 2]]) b = a c = np.concatenate([a, b, a, b], axis=0) d = np.concatenate((a, b), axis=1) print(c) print(d) # %% c # %% c.mean(axis=1) # %% print(config.DATA_MEAN) # %% data = get_trajectory() m = data_mean(data) # %% print(m) print(config.DATA_MEAN) # %% import config from data_processor import standardize_data from data_processor import trim_data from data_reader import get_trajectory from data_analyzer import data_length # %% data = get_trajectory() data[0] # %% data = config.ALL_TRAJECTORY