Interactive Prognostics Dashboard

Engine-level exploration of predictions, uncertainty, and feature attribution.

FD001 test enginesSensor view: last 30 observed cyclesRUL cap: 125 cycles

Engine

Predicted RUL (median)

Interval [--, --]

True RUL

Ground-truth value for selected engine

Confidence

Width -- cycles

Coverage status

Outside interval

Global coverage 83.0%

sensor_measurement_11 - mean ± std across engines

Degradation trend: mean ± std vs RUL (pooled test engines).

No per-engine sensor series found in /public/data/engine_series yet.

SHAP explanation

Local SHAP file not found in /public/data/shap_local; showing global ranking fallback.

Correlation of sensors with RUL

Pooled over cycles and engines · smoothed sensors only

Engine Phase Split and Capping Rationale

RUL trajectory (uncapped)

RUL trajectory (capped at 125)

Selected engine phase share (of total cycles)

Healthy 0.0%

Critical 0.0%

RUL capping (125 cycles) mitigates long healthy-phase dominance so training focuses more on degradation-critical behavior.

Quantile interval chart (across engines)

Showing true RUL, median, and low/high quantile lines.