SignalQualityEvents

Data-quality checks for one numeric signal: gaps against an expected sampling interval, sampling regularity, values outside a physical range, and completeness per window. For gap summaries, coverage and interpolation candidates see DataGapAnalysisEvents.

ts_shape.events.quality.signal_quality · Quality guide · API reference

Example

import ts_shape
from ts_shape.events.quality.signal_quality import SignalQualityEvents

df = ts_shape.make_timeseries(["tank_level"], n_points=7200, freq="1s", baseline=50.0, noise=20.0)
df = df.drop(index=range(3000, 3100))  # 100 s outage

sq = SignalQualityEvents(df, signal_uuid="tank_level")
result = sq.detect_missing_data(expected_interval="1s", tolerance_factor=2.0)
print(result[["start", "gap_duration", "expected_samples_missing"]])
out_of_range = sq.detect_out_of_range(min_value=0.0, max_value=100.0)
completeness = sq.data_completeness(window="1h", expected_interval="1s")

Methods

Method	Returns
detect_missing_data(expected_interval='1s', tolerance_factor=2.0)	gaps longer than tolerance_factor × expected_interval: start, end, gap_duration, expected_samples_missing
sampling_regularity(window='1h')	per window: mean_interval, std_interval, min_interval, max_interval, regularity_score
detect_out_of_range(min_value, max_value, interval_end='next_change')	intervals outside the range: start, end, duration, min_observed, max_observed, direction
data_completeness(window='1h', expected_interval='1s')	per window: expected_count, actual_count, completeness_pct

Notes

• An out-of-range interval ends at the next sample back in range by default, so change-only signals report the real time out of range. interval_end="last_sample" ends it at the last flagged sample.
• For historian data that is compressed or batched, raise tolerance_factor (e.g. to 3) to avoid false gaps.