Skip to content

StatisticalProcessControlRuleBased¤

Applies SPC rules (Western Electric Rules) to a DataFrame for event detection using control limit UUIDs and actual value UUIDs.

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

When to Use¤

Use for real-time quality monitoring when you have a process signal and corresponding control limits. Ideal for detecting shifts, trends, and patterns that indicate a process is going out of control per the 8 Western Electric rules. Requires a DataFrame with actual values and tolerance/control limit columns.

Quick Example¤

from ts_shape.events.quality.statistical_process_control import StatisticalProcessControlRuleBased

spc = StatisticalProcessControlRuleBased(
    df,
    actual_uuid="sensor:temp",
    tolerance_uuid="limit:temp",
)

# Run the full SPC pipeline (calculates limits + applies all rules)
result = spc.process()

# Or apply specific rules only
limits = spc.calculate_control_limits()
violations = spc.apply_rules_vectorized(selected_rules=[1, 2, 5])
summary = spc.interpret_violations(violations)

Key Methods¤

Method Purpose Returns
calculate_control_limits() Calculate mean and 1/2/3 sigma limits DataFrame with control limits
calculate_dynamic_control_limits(method='moving_range', window=20) Adaptive control limits for non-stationary processes DataFrame with dynamic limits
apply_rules_vectorized(selected_rules=None) Apply selected Western Electric rules efficiently DataFrame with violation flags
detect_cusum_shifts(target, k, h) CUSUM shift detection for small sustained shifts DataFrame with CUSUM statistics
interpret_violations(violations_df) Generate human-readable violation interpretations DataFrame with descriptions
process(selected_rules=None) Full SPC pipeline: limits + rules + interpretation DataFrame with complete results
rule_1 through rule_8 Individual Western Electric rule checks DataFrame with single rule flags

Tips & Notes¤

Start with the process() method

The process() method runs the complete pipeline and is the easiest entry point. Use individual rules or apply_rules_vectorized(selected_rules=...) only when you need to tune which rules apply to your specific process.

Related modules

See Also¤