Table 2 Summary of analytical results... | Emerald Publishing

Table 2

Summary of analytical results for edge AI-Based predictive maintenance model

Analysis type	Variable(s) involved	Method applied	Key output/Metric	Figure ref.
Time Series Monitoring	Energy, Vibration, Current_A, Temp_C	Real-Time Data Capture	6-week series, 15-s intervals	Figure 2
Correlation Analysis	Energy vs all sensor variables	Pearson Correlation	Max r = 0.84 (Current_A)	Figure 3
ANOVA	Energy Consumption across time segments	One-Way ANOVA	Significant variation across shifts (p < 0.01)	Figure 4
Prediction Error Distribution	Residuals	Difference (Actual - Predicted)	Error spread visualized	Figure 5
Histogram of Residuals	Residuals	Frequency Count	Approx. Normal distribution of errors	Figure 6
ROC Curve	Energy Level (Binary Classification)	ROC/AUC	AUC = 0.91	Figure 7
ICE Plot	Current_A vs Predicted Energy	Individual Conditional Expectation	Variable-specific impact trajectories	Figure 8

Analysis type	Variable(s) involved	Method applied	Key output/Metric	Figure ref.
Time Series Monitoring	Energy, Vibration, Current_A, Temp_C	Real-Time Data Capture	6-week series, 15-s intervals	Figure 2
Correlation Analysis	Energy vs all sensor variables	Pearson Correlation	Max r = 0.84 (Current_A)	Figure 3
ANOVA	Energy Consumption across time segments	One-Way ANOVA	Significant variation across shifts (p < 0.01)	Figure 4
Prediction Error Distribution	Residuals	Difference (Actual - Predicted)	Error spread visualized	Figure 5
Histogram of Residuals	Residuals	Frequency Count	Approx. Normal distribution of errors	Figure 6
ROC Curve	Energy Level (Binary Classification)	ROC/AUC	AUC = 0.91	Figure 7
ICE Plot	Current_A vs Predicted Energy	Individual Conditional Expectation	Variable-specific impact trajectories	Figure 8

Source(s): Author’s work