Table 3: Comparison of Model Results using Mean Absolute Deviation for Hazardous Liquid Pipelines
Table 3: Comparison of Model Results using Mean Absolute Deviation for Hazardous Liquid Pipelines
| Model | Description | MAD |
|---|---|---|
| Nave model | Previous year's incident count | 1.0 |
| Mileage alone | Modeled with negative binomial | 1.0 |
| PIPP rank | As presently used by PHMSA | 7.6 |
| PIPP total score | Modeled with negative binomial | 2.0 |
| PIPP input variables | Modeled with negative binomial | 1.1 |
| PIPP output variables (L-scores) | Used to calculate PIPP components and modeled with negative binomial | 3.2 |
| Inherent risk - using pipeline data | Infrastructure variables from annual reports and modeled with negative binomial | 1.0 |
| Performance risk - using inspection data about operators | Inspection results from IM inspections | 1.5 |
| Historical risk - using incident data | Number of past three year's incidents, modeled with orthogonal regression | 1.1 |
SOURCE: U.S. Department of Transportation, Research and Innovative Technology Administration, Bureau of Transportation Statistics, http://www.bts.gov.