Region MTBF
This dashboard is a machine learning-based dashboard that shows the predicted failure rates and the error metrics improvement from Servigistics for all the parts for the current period.
Some key benefits:
• Optimal failure rate calculated for each part at each region
• Comparison of global failure rate and ML based failure rate via MAPE, MAD, RMSE, Forecast Accuracy calculation
• $Global Bias ,$ML Bias and $Delta Bias for getting the high financial impact Parts
 | The Price used to calculate $ values for each Part is average of prices for all locations in a region |
• Dashboard to see summary of all parts including key attributes
• Detailed dashboard to analyze a part at any region
• Feature Importance metrics to see what features are critical for each prediction. This helps in understanding the impact of regional or global features in determining the failure rate for a part for a region.
The widgets on the Region MTBF dashboard are separated into three mini-dashboards:
RMSE
This mini-dashboards helps you understand the improvement in RMSE when ML Failure Rate is used.
Widget | Description |
|---|---|
Global RMSE | The sum of RMSE of all the part and regions when global MTBF/Failure Rate is used for forecasting. |
ML RMSE | The sum of RMSE of all the part and regions when ML MTBF/Failure Rate is used for forecasting. |
Delta RMSE | The difference between Global RMSE and ML RMSE. |
Improvement% | The improvement percentage in RMSE when ML MTBF/Failure Rate is used over global for forecasting. |
Causal Data By Region
This mini-dashboard helps you analyze the causal data by region.
Widget | Description |
|---|---|
# Parts | The total count of parts. |
Part Count By Region | The count of parts by region. |
Causal Value By Region | A pie chart of the causal value by region. |
Part Population By Region | A pie chart of the part population by region. |
Part Summary
This mini-dashboard helps you analyze the causal data by part.
Click a Part Number hyperlink to open the Region MTBF Detail dashboard to view the detailed prediction for the selected part number.