Skip to content

ProcessWindowEvents¤

Analyze time-windowed process statistics for shift reports, SPC context, and trend monitoring.

Module: ts_shape.events.engineering.process_window
Guide: Process Engineering

When to Use¤

Use as a building block for shift-level process monitoring. Answers "how is my process doing this hour/shift?" with windowed statistics and shift detection. Provides the statistical foundation for SPC charts, shift handover reports, and trend dashboards.

Quick Example¤

from ts_shape.events.engineering.process_window import ProcessWindowEvents

analyzer = ProcessWindowEvents(
    df=process_data,
    uuid="mixer_torque_01"
)

# Descriptive statistics per 1-hour window
stats = analyzer.windowed_statistics(window="1h")

# Detect windows where the mean has shifted
shifts = analyzer.detect_mean_shift(window="1h", sensitivity=2.0)

# Detect windows where variance has changed
var_changes = analyzer.detect_variance_change(
    window="1h",
    ratio_threshold=2.0
)

# Compare each window to the overall baseline
comparison = analyzer.window_comparison(window="1h")

Key Methods¤

Method Purpose Returns
windowed_statistics(window='1h') Per-window descriptive stats DataFrame with mean, std, min, max
detect_mean_shift(window='1h', sensitivity=2.0) Mean shift detection DataFrame of shift events
detect_variance_change(window='1h', ratio_threshold=2.0) Variance change detection DataFrame of variance events
window_comparison(window='1h') Compare windows to baseline DataFrame with comparison metrics

Tips & Notes¤

Align windows to shift boundaries

Use window='8h' for 8-hour shifts or window='12h' for 12-hour shifts. This ensures each window maps to exactly one shift, making handover reports straightforward. Combine with detect_mean_shift() to flag shifts that deviate from normal.

Related modules

See Also¤