def load_external_ts(filepath):
"""
Loads space delimited time series text file from disk to be searched on.
Args:
filepath: path to time series file
Returns:
A 100 point interpolated ArrayTimeSeries object for times between 0 and 1.
Notes:
- Only considers the first two columns of the text file (other columns are discarded)
- Only evaluates time values between 0 and 1
- First column is presumed to be times and second column is presumed to be light curve values.
"""
data = load_nparray(filepath)
data = data[:, :2] # truncate to first 2 cols
# Remove rows with duplicate time values (if they exist) and resorts to ensure ts in ascending order
_, indices = np.unique(data[:, 0], return_index=True)
data = data[indices, :]
times, values = data.T
full_ts = ats.ArrayTimeSeries(times=times, values=values)
interpolated_ats = full_ts.interpolate(
np.arange(0.0, 1.0, (1.0 / TS_LENGTH)))
return interpolated_ats
def load_ts(ts_fname):
"""Helper to load previously generated ts file from disk"""
if ts_fname.startswith("ts-"):
filepath = LIGHT_CURVES_DIR + ts_fname
data = load_nparray(filepath)
times, values = data.T
return ats.ArrayTimeSeries(times=times, values=values)
else:
raise ValueError("'%s' does not appear to be a time series file" %
ts_fname)
def load_ts(LIGHT_CURVES_DIR):
"""Loads time series text files; returns dict keyed to filename"""
timeseries_dict = {}
for file in os.listdir(LIGHT_CURVES_DIR):
if file.startswith("ts-") and file.endswith(".txt"):
filepath = LIGHT_CURVES_DIR + file
data = np.loadtxt(filepath)
times, values = data.T
ts = ats.ArrayTimeSeries(times=times, values=values)
timeseries_dict[file] = ts
return timeseries_dict
def random_ts(jitter,length = 100):
"""
Creates uniform random array time series object.
Args:
jitter: how much variation to data points should have. (A value of zero is basically a straight line.)
length: Length of the time series. Defaults to 100 values.
Returns:
An array time series object.
"""
times = np.arange(0.0, 1.0, (1.0 / length))
values = jitter*np.random.random(100)
return ats.ArrayTimeSeries(times=times, values=values)
def tsmaker(mean, scale, jitter, length = 100):
"""
Creates array time series object based on the normal distribution PDF.
Args:
mean: center of distribution
scale: variance/spread of distribution.
jitter: how much uniform-random variation is added on top of the distribution
length: Length of the time series. (Defaults to 100 values)
Returns:
An array time series object.
"""
times = np.arange(0.0, 1.0, (1.0 / length))
values = norm.pdf(times, mean, scale) + jitter*np.random.randn(100)
return ats.ArrayTimeSeries(times=times, values=values)
def standardize(ts):
"""standardize timeseries ts by its mean and std deviation"""
stand_vals = (ts.values() - ts.mean())/ts.std()
return ats.ArrayTimeSeries(times=ts.times(), values=stand_vals)