Roshani Lobo completed Predictive Maintenance: Industrial ML for Fault Detection

Built an end-to-end rolling-bearing fault detector from vibration snapshots: visualized healthy vs faulty signals, identified impulsive fault behavior, selected kurtosis as the strongest time-domain indicator, trained a compact depth-capped decision tree on training bearings only, and evaluated it on held-out bearings with reported catch-rate and false-alarm rate. The maintenance memo defended a 0.62 alarm threshold using explicit cost tradeoffs and correctly localized the fault to the outer race via BPFO.

• Leakage-Free Feature Engineering
• Honest Held-Out Evaluation
• Cost-Based Decision Making
• Interpreting Defect Frequencies