12 · Python · Machine learning
Predictive maintenance
Maintenance follows the calendar, not the measurement.
Compare fixed running hours with condition-based maintenance, including precision and recall.
Demo on fictional data — never customer data.
Precision and recall are counted on all 1,600 installations, with a fixed label per installation. The chart shows the selected series.
- Risk
- 86.0%
- Failure in week
- 19
- Precision
- 76.8%
- Recall
- 96.6%
- False positives
- 111
Vibration by week
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
Hour-based maintenance arrives in week 15. Condition-based maintenance arrives in week 14.
precision = true_positive / (true_positive + false_positive)
recall = true_positive / (true_positive + false_negative)