|Methods for Forecasting Attrition and Force Strength: Quantifying and Reporting Uncertainty for Managing Risk in Military Personnel Planning|
|System Analysis and Studies|
Attrition, Forecasting, Military, Personnel, Planning, Prediction Intervals, Risk, SAS, Uncertainty
Attrition forecasts often report only a single attrition value. This can be misleading because the historical data and method used can introduce uncertainties into the process. Consequently, prediction limits must be identified, computed, and represented. A representation that is too simplistic can produce an exaggerated and false sense of accuracy in the forecast and, in cases where the horizon of the forecasted is large, the plausibility of the forecasted values is questionable. The proper quantification of prediction accuracy, minimization the adverse effect of uncertainty in prediction, and, finally, level of detail (prediction intervals derived from the inherited uncertainty) in the visual and numerical representation of the forecasted data will all have a significant positive impact on the success of personnel management policies and strategies. Accordingly, this activity aims to determine how much information should be conveyed in a report to allow for a proper understanding and use of attrition forecasts. Specifically, this activity aims to identify the forecasting methods that are (i) suitable for specific attrition problems, (ii) have a small error, and (iii) provide an output (i.e., a set of values) that, with appropriate representation, can be easily interpreted by personnel planners.
This proposal has the following objectives:
(i) build an inventory of definitions of attrition (e.g., computation rules, how the definition is used in planning) with advantages and disadvantages of each definition;
(ii) build an annotated inventory of the methods now used to forecast attrition and total force strength;
(iii) develop new methods for computing prediction intervals for forecasting; and
(iv) build an inventory of styles used to report forecasts and projections.
From this body of knowledge we will extract best practices for definitions, methods, and reporting styles for attrition forecasts. We expect to propose a standard for presenting forecasted attrition and total force strength that can be used by NATO and partner nations.
(i) Computational methods for personnel attrition;
(ii) methods for forecasting attrition and total force strength;
(iii) computing prediction intervals for forecasting; and
(iv) quantifying uncertainty in forecast projections.