Lifetime Estimation Based on Warranty Data Without Date-of-Sale Information-Cases Where Usage Time Distributions Are Known

DOI Web Site 3 Citations 9 References
  • WANG Lianhua
    The Graduate School of Information Systems, The University of Electro-Communications
  • SUZUKI Kazuyuki
    The Graduate School of Information Systems, The University of Electro-Communications

Bibliographic Information

Other Title
  • 販売時点情報が失われた保証期間中の故障データのみに基づく寿命推定法－使用頻度分布が既知の場合－

Search this article

Abstract

In field reliability research, concern over warranty claim data has been growing in recent years. Warranty claim records, which are usually provided by a maintenance or finance department, are often grouped according to the month when the maintenance or accounting is carried out, while sales numbers obtained from a sales department are usually aggregated according to sales month. In this case, the actual sales date of a failed product is unknown. This paper discusses nonparametric estimation of lifetime distributions from warranty data in this context. We concentrate on those products for which 1ifetimes are measured in usage time, not in calendar time, and derive a nonparametric MLE(maximum likelihood estimator)based on the concept of subdistribution in cases where usage time distributions are known. Conducting simulation experiments to compare this estimator with the MLE for the case where sales dates are available, we find that the two estimators have almost the same level of precision. This indicates that the date-of-sale information is not very important for estimating lifetimes when lifetimes are measured in usage time. We obtained similar results when we extended the method to the case of competing failure modes.

Journal

Citations (3)*help

See more

References(9)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top