AN EMPIRICAL TEST OF EXPONENTIAL SMOOTHING

Abstract

This Memorandum, a comparative study of techniques for predicting the demand for spare parts, attempted to discover the potential advantages which exponential smoothing has over the moving average procedures the Air Force is now using. Various forms of exponential smoothing and averaging were applied to three sample sets of data: Hi-Valu and Category II recoverable B-52 parts, components of the Falcon missile, and Category III depot issues of B-52 items. The study led to the following findings and conclusions: (1) For any of the three sets of data, exponential smoothing was not a significantly better prediction technique than the cumulative issue rate procedures now being used in the Air Force. Nevertheless, it does have definite computational advantages which may be valuable. (2) A measure of aggregate loss, such as the loss function introduced in this study, should be used to select preferred smoothing techniques. (3) The use of program element information for the Falcon components improved the accuracy of our predictions; application of requisition data, which was available for the Category III items, did not. (4) With any of the techniques applied to the Category III items, prediction accuracy did not increase substantially when the base period was made longer than one year.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1964
Accession Number
AD0600239

Entities

People

  • Craig C. Sherbrooke
  • Max Astrachan

Organizations

  • RAND Corporation

Tags

Communities of Interest

  • Air Platforms
  • Weapons Technologies

DTIC Thesaurus Topics

  • Accuracy
  • Air Force
  • Aircrafts
  • Control
  • Corporations
  • Data Processing
  • Data Science
  • Data Storage Systems
  • Information Science
  • Measuring Instruments
  • Operations Research
  • Probability
  • Probability Distributions
  • Spare Parts
  • Statistical Analysis
  • Statistical Decision Theory
  • Statistical Tests

Readers

  • Business Analytics
  • Logistics and Supply Chain Management.
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks