Self-directed project
Predictive Maintenance: Industrial ML for Fault Detection
Condition-Monitoring Engineer · AI / ML · July 2026
A ProoV case study · educational project, not employment
Built an end-to-end rolling-bearing fault detector from vibration data: engineered time-domain and envelope-spectrum features, trained a depth-capped decision tree on training bearings only, and evaluated it on a held-out set with explicit catch-rate and false-alarm metrics. The final memo defended a 0.40 alarm threshold using the student’s own cost numbers and correctly localized the fault to the outer race via BPFO.
Graded against
- Feature engineering (signal to honest numbers)25%
- Held-out evaluation (honest measurement)25%
- Cost-based threshold decision + memo25%
- Compact model + embedded-budget justification15%
- Plain-language explanation10%
Passed · pass mark 60/100
What stood out6
- Used kurtosis as the root split and stated a shallow tree depth of 1 with 3 nodes, which fits the embedded-budget constraint.
- Reported held-out catch-rate and false-alarm rate explicitly and tied the threshold choice to miss-vs-inspection costs.
- Correctly mapped the dominant defect frequency to BPFO and the outer race in the maintenance memo.
- Leakage-Free Feature Engineering
- Honest Held-Out Evaluation
- Physics-Based Fault Localization
The work I submitted19 tasks
11 tasks · 6.4k characters · Python · 8 written answers · 437 words
- Initial Prediction50 characters
- Load Plot Signal205 characters
- Load Plot Signal CodePython · 1.8k characters
- Example Rms WorkedPython · 697 characters
- Time Features143 characters
- Freq FeaturesPython · 1.3k characters
- Fault Location239 characters
- Train Classifier363 characters
- Evaluation Heldout288 characters
- Threshold Decision334 characters
- Decision Memo1.0k characters
- Initial Prediction6 words
- Load Plot Signal15 words
- Load Plot Signal Code163 words
- Example Rms Worked72 words
- Time Features10 words
- Freq Features126 words
- Fault Location14 words
- Train Classifier31 words
Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.