...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Quality Control and Applied Statistics >The cost of errors in software development: Evidence from industry
【24h】

The cost of errors in software development: Evidence from industry

机译：软件开发中的错误代价：来自行业的证据

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

This research develops a model for costs generated in software development focusing on cost associated with software error correction and detection. The paper is organized into the following sections. Section 1 discusses the software life cycle costs. Section 2 of the paper discusses the data collection and control for bias for the empirical study. Section 3 presents a series of pairwise specification searches that describe the influence of particular cost drivers on cost. Section 4 synthesizes a multi-parameter cost model. Section 5 discusses the implication for the software industry and suggests avenues for extending the research. With regard to data collection, data was collected from a packaged software vendor with global operations and a number of competitive products. The vendor releases the software to 31 strategically important global markets after translating the software to the local language, culture and practice. The vendor observed that the cost of translation to the local language contributed to less than 10 percent of total cost incurred in the localization phase of development. Figure 1 in the text shows the process of error testing discovery and acquisition. With regard to error cost, the following was observed. The vendor's internal study concluded that the largest single contributor to the localization cost was the detection and correction of errors. The author provides two hypotheses for the same. The first hypothesis is that the error count is exponentially distributed in time and with c.d.f. and the second hypothesis is that the error rate is distributed according to some other continuous, monotonic increasing c.d.f. To test the first hypothesis, the observed error count was reexpressed via a Box-Cox transformation for various components. Table 1 in the article summarizes the findings. On specific searches, the author has carried out statistical analysis that tested for multi-collinearity. On regression of error occurrence and software cost, errors are prioritized from 1 to 3. A serious error is priority 1. A moderately severe error is priority 2, and a minor error is priority 3. When priority 3 errors are removed, regression results improve significantly. Priority 1 errors are likely to need immediate attention. On regression of error resolution and software cost, the author states that most of the cost arises from errors that require some system redesign. This is substantiated from the regression results given in the article. With regard to regression of life cycle phase of an error and software cost, the author concludes that errors from the earlier phase are much more influential on costs than those for the later phase. This is supported by the regression results given in the article. The author concludes the paper with implications for the software industry. (21 refs.)
机译：这项研究针对软件开发中产生的成本开发了一个模型，重点是与软件错误纠正和检测相关的成本。本文分为以下几节。第1节讨论了软件生命周期成本。本文的第二部分讨论了用于实证研究的偏见的数据收集和控制。第3节介绍了一系列成对的规范搜索，描述了特定成本动因对成本的影响。第4节综合了多参数成本模型。第五部分讨论了对软件行业的影响，并提出了扩大研究范围的途径。关于数据收集，数据是从具有全球运营和众多竞争产品的打包软件供应商那里收集的。在将软件翻译成当地语言，文化和惯例后，供应商将软件发布到31个具有战略意义的全球市场。供应商注意到，翻译成本地语言的成本只占开发本地化阶段总成本的不到10％。文本中的图1显示了错误测试发现和获取的过程。关于错误成本，观察到以下情况。供应商的内部研究得出的结论是，本地化成本最大的单一因素是错误的检测和纠正。作者提供了两个相同的假设。第一个假设是错误计数随时间以c.d.f指数分布。第二个假设是错误率是根据其他连续的单调递增c.d.f分布的。为了检验第一个假设，通过Box-Cox变换针对各种组件重新表达了观察到的错误计数。本文的表1总结了这些发现。在特定搜索中，作者进行了统计分析，以测试多重共线性。对于错误发生和软件成本的回归，将错误的优先级从1到3。优先级为1的严重错误。中等优先级的错误为2的优先级，次要错误为3的优先级。删除优先级3的错误后，回归结果将得到改善显着。优先级为1的错误可能需要立即关注。关于错误解决和软件成本的回归，作者指出，大部分成本来自需要重新设计系统的错误。这可以从文章中给出的回归结果中得到证实。关于错误的生命周期阶段和软件成本的回归，作者得出结论，早期阶段的错误对成本的影响要比后期阶段的错误大得多。本文提供的回归结果对此提供了支持。作者的结论对软件行业产生了影响。 （21篇）

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号