← Back to the chapter
Technical detail · Data Preprocessing Workflow
Under the Hood: audit the HVAC cost estimate
Quantifying the 1-10-100 Rule for IoT Data Quality
Use this arithmetic when reviewing the cost assumptions behind a prevention decision.
The classic 1-10-100 quality-cost rule of thumb illustrates $1 to prevent, $10 to correct, and $100 when a problem causes failure. ASQ
Example: Smart HVAC with stuck sensor reading 15°C in summer
Prevention Cost ($1 equivalent - timestamp staleness check):
CPU Cost=1 comparison/reading×10−6 seconds×3600 seconds$0.001
≈2.8×10−13 per reading≈$0 (negligible)
Correction Cost ($10 equivalent - retrospective data repair):
Storage Scan=86,400 readings/day×30 days×10−4 s/scan
=259 seconds compute=$0.07 cloud compute
+Engineer Time=2 hours×$100/hour=$200
Failure Cost ($100 equivalent - heater ran 8 hours unnecessarily):
Energy Waste=5 kW×8 hours×$0.12/kWh=$4.80/day
×30 days before detected=$144
Total Failure Cost=$144+$200 (investigation)=$344
Actual Ratio:
PreventionCorrection=$0$200=undefinedPreventionFailure=$0$344=undefined
The prevention step is inexpensive in this example, but a ratio requires a nonzero, consistently measured prevention cost; choose edge validation based on the measured cost and risk of failures.
← Back to the chapter