Skip to content

CapabilityTrendingEvents¤

Track process capability indices (Cp, Cpk, Pp, Ppk) over rolling time windows to detect capability degradation before quality escapes occur.

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

When to Use¤

Use for trending process capability over time. Detects capability degradation before it results in quality escapes. Set up as a daily dashboard to track Cpk trends and predict when intervention is needed. Requires process data with defined specification limits and sufficient sample sizes per window.

Quick Example¤

from ts_shape.events.quality.capability_trending import CapabilityTrendingEvents

cap = CapabilityTrendingEvents(
    df,
    value_column="value_double",
    usl=105.0,
    lsl=95.0,
)

# Compute capability indices per 8-hour window
trend = cap.capability_over_time(window="8h")

# Alert when Cpk drops below 1.33
drops = cap.detect_capability_drop(window="8h", min_cpk=1.33)

# Forecast when Cpk will breach the threshold
forecast = cap.capability_forecast(window="8h", horizon=5, threshold=1.33)
print(f"Predicted breach in {forecast['windows_to_breach']} windows")

# Estimate yield, DPMO, and sigma level
yield_stats = cap.yield_estimate(window="8h")

Key Methods¤

Method Purpose Returns
capability_over_time(window='8h') Compute Cp, Cpk, Pp, Ppk per rolling window DataFrame with capability indices per window
detect_capability_drop(window='8h', min_cpk=1.33) Flag windows where Cpk falls below threshold DataFrame with drop events
capability_forecast(window='8h', horizon=5, threshold=1.33) Predict future Cpk breach based on trend Dictionary with forecast metrics
yield_estimate(window='8h') Estimate yield percentage, DPMO, and sigma level DataFrame with yield metrics per window

Tips & Notes¤

Use min_cpk=1.33 as the default threshold

A Cpk of 1.33 corresponds to a 4-sigma process (~63 DPMO). For safety-critical processes, raise this to 1.67 (5-sigma). Set detect_capability_drop to alert at your threshold so you can intervene before parts go out of spec.

Related modules

  • Tolerance Deviation — point-in-time capability indices and severity classification
  • SPC Rules — Western Electric rule violations that often precede capability drops
  • Sensor Drift — sensor drift as a root cause of apparent capability degradation

See Also¤