Modelling and Quantitative Cost Risk Analysis:

Results Are Incorrect or Unexpected

 

Overview

This article addresses scenarios where Predict! Risk Analyser produces unexpected or incorrect results during Quantitative Cost Risk Analysis.

 

Issues:

• Confidence levels (for example, 80%) not aligning with expectations
• Analysis referencing incorrect cost or schedule data
• Results varying between analysis runs

These outcomes typically indicate issues related to model configuration, input data, or interpretation of probabilistic outputs.

 

Solution

1. Confirm model references
   • Verify that the correct cost or schedule cells are used in the model.
   • Ensure no unintended ranges or incorrect mappings are included.
2. Validate uncertainty inputs
   • Review minimum, most likely, and maximum values.
   • Confirm probability distributions are correctly assigned.
3. Verify risk linkages
   1. Ensure risks are linked to the appropriate cost or schedule elements.
   2. Check for missing or duplicate risk associations.
4. Review confidence level interpretation
   • Confirm that percentile outputs are interpreted correctly.
   • Example: An 80% confidence level represents a value with an 80% probability of not being exceeded.
5. Re-run the analysis
   • Execute the analysis after adjustments.
   • Confirm results are consistent across multiple runs.

 

Troubleshooting

• Confidence values appear too high or too low:
  • Review uncertainty ranges and distribution types.
• Incorrect data is referenced:
  • Validate cell mappings and imported datasets.
• Results differ between runs:
  • Confirm no model inputs are changing between executions.
  • Check for external or dynamically updated data sources.

 

Best Practices

• Validate models using simple baseline scenarios before increasing complexity.
• Maintain consistent naming conventions for cost and schedule inputs.
• Document modelling assumptions and data sources.
• Periodically review uncertainty inputs and risk mappings.