Skip to content

SignalQualityEvents¤

Detect data quality issues in a numeric signal: missing data, irregular sampling, out-of-range values, and data completeness.

Module: ts_shape.events.quality.signal_quality
Guide: Quality Control & SPC

When to Use¤

Use to validate data quality before analysis. Check for gaps in historian data, irregular sampling from PLCs, or sensor failures producing out-of-range values. Run this as a first step in any analytics pipeline to avoid drawing conclusions from incomplete or corrupted data.

Quick Example¤

from ts_shape.events.quality.signal_quality import SignalQualityEvents

sq = SignalQualityEvents(df, value_column="value_double")

# Find gaps where expected samples are missing
gaps = sq.detect_missing_data(expected_freq="1s", tolerance_factor=2.0)

# Check sampling regularity per hour
regularity = sq.sampling_regularity(window="1h")

# Flag physically impossible readings
out_of_range = sq.detect_out_of_range(min_value=0.0, max_value=100.0)

# Compute hourly completeness percentage
completeness = sq.data_completeness(window="1h", expected_freq="1s")

Key Methods¤

Method Purpose Returns
detect_missing_data(expected_freq='1s', tolerance_factor=2.0) Find gaps exceeding expected sampling interval DataFrame with gap start/end and duration
sampling_regularity(window='1h') Compute inter-sample interval statistics per window DataFrame with mean, std, min, max intervals
detect_out_of_range(min_value, max_value) Flag values outside physical or expected range DataFrame with out-of-range flags
data_completeness(window='1h', expected_freq='1s') Calculate percentage of expected samples received DataFrame with completeness percentages

Tips & Notes¤

Set tolerance_factor based on your data source

For PLC data with a strict 1-second cycle, a tolerance_factor=2.0 works well. For historian data that may batch-compress, increase to 3.0 or higher to avoid false gap detections.

Related modules

See Also¤