What hazami demonstrated

Built a compact rolling-bearing fault detector end to end: 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 catch-rate and false-alarm rate. The work shows practical judgment in choosing a low-compute model and defending an operating threshold with measured trade-offs.

• Leakage-Free Feature Engineering
• Honest Model Evaluation
• Compact Interpretable Modeling
• Cost-Based Operational Decision-Making