def test_robustness_no_ts(self):
"""
No time series is provided
Check that the result is empty
"""
# Prepare inputs
ts_list = []
# Prepare expected output
expected_results = {}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=2000,
timestamp_position="BEG",
aggregation_method="AVG",
nb_points_by_chunk=4,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_empty_chunks(self):
"""
Compute the downsampling on multiple time series having many points not evenly distributed
Check the downsampling works with empty chunks
"""
# Prepare inputs
ts_list = [gen_ts(3)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12, 5.5], [1e12 + 2000, 7.0],
[1e12 + 100000, 7.0], [1e12 + 102000, 7.5]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=2000,
timestamp_position="BEG",
aggregation_method="AVG",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_single_point(self):
"""
Compute the downsampling on time series having 1 point
Check the new time series is the same as the original one
"""
# Prepare inputs
ts_list = [gen_ts(4)]
# Prepare expected output
expected_results = {ts_list[0]['tsuid']: [[1e12, 42.0]]}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=2000,
timestamp_position="BEG",
aggregation_method="MAX",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_alignment_end(self):
"""
Compute the downsampling on a single time series using END alignment
Check the time series is processed and result matches the alignment method
"""
# Prepare inputs
ts_list = [gen_ts(1)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12 + 6000, 8.0], [1e12 + 12000, 42.0],
[1e12 + 18000, 8.0]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=6000,
timestamp_position="END",
aggregation_method="MAX",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_aggregation_med(self):
"""
Compute the downsampling on a single time series using MED aggregation (median)
To compute the median, sort (ascending) the values in the desired period and take the middle point
(or apply a linear interpolation in case of even values)
Check the time series is processed and result matches the aggregation method
"""
# Prepare inputs
ts_list = [gen_ts(1)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12, 5.5], [1e12 + 10000, 8.0]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=10000,
timestamp_position="BEG",
aggregation_method="MED",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_aggregation_min(self):
"""
Compute the downsampling on a single time series using MIN aggregation
Check the time series is processed and result matches the aggregation method
"""
# Prepare inputs
ts_list = [gen_ts(1)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12, 5.0], [1e12 + 2000, 6.0],
[1e12 + 4000, -15.0], [1e12 + 6000, 3.0],
[1e12 + 8000, 2.0], [1e12 + 10000, 8.0],
[1e12 + 12000, 8.0]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=2000,
timestamp_position="BEG",
aggregation_method="MIN",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_not_aligned(self):
"""
Compute the downsampling on a single time series with a resampling period not a multiple of the original period
Check the time series is processed
"""
# Prepare inputs
ts_list = [gen_ts(1)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12, 6.0], [1e12 + 1400, 6.0],
[1e12 + 2800, 8.0], [1e12 + 4200, 2.0],
[1e12 + 5600, 6.0], [1e12 + 7000, 3.0],
[1e12 + 8400, 42.0], [1e12 + 9800, 8.0],
[1e12 + 11200, 8.0], [1e12 + 12600, 8.0]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=1400,
timestamp_position="BEG",
aggregation_method="MAX",
nb_points_by_chunk=4,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)
def test_multiple_ts(self):
"""
Compute the downsampling on multiple time series without any constraint
Check all time series are processed
"""
# Prepare inputs
ts_list = [gen_ts(1), gen_ts(2)]
# Prepare expected output
expected_results = {
ts_list[0]['tsuid']: [[1e12, 5.5], [1e12 + 2000, 7.0],
[1e12 + 4000, -6.5], [1e12 + 6000, 4.5],
[1e12 + 8000, 22.0], [1e12 + 10000, 8.0],
[1e12 + 12000, 8.0]],
ts_list[1]['tsuid']: [[1e12, 5.5], [1e12 + 2000, 7.0],
[1e12 + 4000, -6.5], [1e12 + 6000, 4.5],
[1e12 + 8000, 3.5], [1e12 + 12000, 10.0]]
}
result = None
try:
# Call algorithm
result = downsampling_ts(ts_list=ts_list,
resampling_period=2000,
timestamp_position="BEG",
aggregation_method="AVG",
nb_points_by_chunk=50000,
generate_metadata=True)
# Check the results
self._check_results(ts_list, result, expected_results)
finally:
# Cleanup
self._cleanup_ts(result)