# PCOEx Decision Summary: Total Revenue

Decision support only. Recommendations do not automatically change operations.

## Model Fit

- R2: 0.99157
- Cross-validated R2: 0.904585
- RMSE: 10994.617081
- Cross-validated RMSE: 36988.648918
- MAE: 8066.514153
- Samples: 30
- Confidence score: 0.699267

## Literature Visuals

- PLS model selection plot: response error by component count.
- PLS response plot: observed response values against predicted response values.
- PLS residual plot: residuals against predicted response values.
- PLS coefficient table: top predictors by standardized coefficient magnitude.
- Heatmap canonical covariates PCOEx-PMS: PMEI category canonical correlations against the selected response.

## Literature Workflow

- Data validation: Import contract, missing-value checks, zero-variance exclusion
  - Status: completed
  - Result: Loaded 30 usable periods, quarantined 0 invalid observations, and excluded 0 non-informative predictors.
- Seasonality encoding: One-hot month features
  - Status: completed
  - Result: Added month indicators so recurring seasonal effects do not masquerade as operational levers.
- PLS regression: Partial Least Squares, single-target response
  - Status: completed
  - Result: Fit the target with R2 0.992; cross-validation returned R2 0.905, RMSE 36988.649, and MAE 25948.212.
- Forward selection regression: Greedy predictor entry by incremental R2
  - Status: completed
  - Result: Selected predictors step-by-step to compare the PLS ranking against a simpler regression path.
- Canonical category validation: Canonical Correlation Analysis by PMEI category
  - Status: completed
  - Result: Calculated canonical category covariates for people, material, equipment, and information categories against the selected response variable.
- Bootstrap stability: Repeated resampling of PLS top predictors
  - Status: completed
  - Result: The top action vector appeared in 5% of bootstrap top-predictor sets.
- Predictive model comparison: Cross-validated PLS, forward-selection linear, time-series baseline, and tree ensemble
  - Status: completed
  - Result: Selected Forward-selection linear by lowest cross-validated RMSE.
- Action vector ranking: PLS variable importance and coefficient direction, with CCA category validation
  - Status: completed
  - Result: Ranked 10 actionable leading indicators by PLS importance; top category-validation score is 1.000.

## Ranked Action Vectors

1. work_in_process
   - Importance: 0.275236
   - Practical label: reduce
   - Statistical interpretation: negative coefficient
   - Confidence: low
   - PLS importance: 0.220604
   - PLS coefficient: -565.440992
   - CCA-style category validation: 0.999996
   - Action: Prioritize a focused improvement project to reduce Work-in-Process and review Total Revenue in the next analysis cycle.
2. maintenance_time_spent
   - Importance: 0.250139
   - Practical label: reduce
   - Statistical interpretation: negative coefficient
   - Confidence: high
   - PLS importance: 0.200489
   - PLS coefficient: -4673.023972
   - CCA-style category validation: 0.532064
   - Action: Prioritize a focused improvement project to reduce Maintenance Time Spent and review Total Revenue in the next analysis cycle.
3. breakdown_regular_hours
   - Importance: 0.119351
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: medium
   - PLS importance: 0.095661
   - PLS coefficient: 1197.084668
   - CCA-style category validation: 0.532064
   - Action: Monitor Breakdown Regular Hours as a leading signal for Total Revenue; use local operational context before launching work.
4. maintenance_occurrences
   - Importance: 0.105528
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: high
   - PLS importance: 0.084582
   - PLS coefficient: 4242.408826
   - CCA-style category validation: 0.532064
   - Action: Monitor Maintenance Occurrences as a leading signal for Total Revenue; use local operational context before launching work.
5. repair_occurrences
   - Importance: 0.084839
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: high
   - PLS importance: 0.067999
   - PLS coefficient: 4182.809855
   - CCA-style category validation: 0.532064
   - Action: Monitor Repair Occurrences as a leading signal for Total Revenue; use local operational context before launching work.
6. repair_time_spent
   - Importance: 0.077556
   - Practical label: reduce
   - Statistical interpretation: negative coefficient
   - Confidence: high
   - PLS importance: 0.062162
   - PLS coefficient: -1640.440087
   - CCA-style category validation: 0.532064
   - Action: Prioritize a focused improvement project to reduce Repair Time Spent and review Total Revenue in the next analysis cycle.
7. recut_occurrences
   - Importance: 0.053639
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: medium
   - PLS importance: 0.042992
   - PLS coefficient: 1619.061344
   - CCA-style category validation: 0.223983
   - Action: Monitor Recut Occurrences as a leading signal for Total Revenue; use local operational context before launching work.
8. recut_time_spent
   - Importance: 0.026419
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: low
   - PLS importance: 0.021175
   - PLS coefficient: 636.66247
   - CCA-style category validation: 0.223983
   - Action: Monitor Recut Time Spent as a leading signal for Total Revenue; use local operational context before launching work.
9. warranty_return_authorizations
   - Importance: 0.006603
   - Practical label: reduce
   - Statistical interpretation: negative coefficient
   - Confidence: high
   - PLS importance: 0.005292
   - PLS coefficient: -306.085067
   - CCA-style category validation: 0.999996
   - Action: Prioritize a focused improvement project to reduce Warranty Return Authorizations and review Total Revenue in the next analysis cycle.
10. warranty_shipments
   - Importance: 0.00069
   - Practical label: monitor
   - Statistical interpretation: positive coefficient; operational review
   - Confidence: low
   - PLS importance: 0.000553
   - PLS coefficient: 26.812094
   - CCA-style category validation: 0.999996
   - Action: Monitor Warranty Shipments as a leading signal for Total Revenue; use local operational context before launching work.