Statistical Engineering: Statistical Methods for Process Improvement and Decision Making

统计工程：过程改进和决策的统计方法

• 批准号: RGPIN-2017-04062
• 负责人: Steiner, Stefan
• 金额: $ 2.48万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2020
• 资助国家: 加拿大
• 起止时间: 2020-01-01 至 2021-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=709372
• 关键词: Statistical; Engineering; Methods; Process; Improvement

The general goals of this research include improving efficiency, reducing costs and better decision making and allocation of resources in business, industrial and medical applications. More specifically, I propose to pursue the following three related research themes: estimating current process performance from streaming data, augmenting the design and analysis of experiments using baseline data and improving the assessment of measurement systems. There are many situations where we obtain data at regular time intervals and wish to estimate the current process performance. For example, we may run weekly customer satisfaction surveys. Here we expect satisfaction to change slowly over time as we address customer concerns, develop new products, react to actions by our competitors, etc. This research aims to develop robust methods for assessing the current process condition that takes into account sampling variation, possible changes over time and the effect of covariates. New methods that combine data in a sensible way can greatly improve decision making. There are many related application areas, such as monitoring hospital or medical lab performance. Experiments, where we deliberately change a process to assess the effect of the change, are widely used in virtually all scientific inquiry. In this research I focus specifically on the use of experiments in improvement projects for existing processes. In this context, we usually have a lot of available observational “baseline” data. The goals of this project are to show how to best use this available process data to improve the planning of experiments and also to quantify the gains in efficiency resulting from incorporating the baseline data in the analysis. Since experiments are typically expensive and disruptive to run, such improvements would save time and money. Measurement systems are essential in virtually all areas of scientific inquiry. Without precise and well calibrated measurement systems process understanding and/or improvement are difficult to achieve. This research focuses on improving the assessment of binary measurement systems that are essential as inspection systems in industry and as diagnostic tests in medicine. Efficient and effective measurement system assessments and comparisons would enhance our ability to make informed decisions about which measurement systems to use and how to improve existing measurement systems.

这项研究的总体目标包括提高效率，降低成本，以及在商业、工业和医疗应用中更好地做出决策和分配资源。更具体地说，我建议进行以下三个相关的研究主题：从流数据估计当前过程性能，使用基线数据增强实验的设计和分析，以及改进测量系统的评估。 在许多情况下，我们以固定的时间间隔获取数据，并希望估计当前的流程性能。例如，我们可能每周进行一次客户满意度调查。在这里，我们预计满意度将随着时间的推移而缓慢变化，因为我们解决了客户的担忧、开发新产品、对竞争对手的行动做出反应等。这项研究旨在开发稳健的方法来评估当前的工艺条件，其中考虑到抽样变化、可能的时间变化和协变量的影响。以合理的方式组合数据的新方法可以极大地改善决策。有许多相关的应用领域，例如监控医院或医学实验室的表现。 我们故意改变一个过程以评估改变的效果的实验，在几乎所有的科学研究中都被广泛使用。在这项研究中，我特别关注在现有流程的改进项目中使用实验。在这种情况下，我们通常有大量可用的观测“基线”数据。该项目的目标是展示如何最好地利用这些现有的过程数据来改进实验计划，并量化将基线数据纳入分析所带来的效率收益。由于实验通常是昂贵的，而且会造成破坏，这样的改进将节省时间和金钱。 测量系统在几乎所有的科学研究领域都是必不可少的。如果没有精确和良好校准的测量系统，就很难理解和/或改进过程。这项研究的重点是改进对二进制测量系统的评估，这些系统在工业上作为检查系统，在医学上作为诊断测试是必不可少的。高效和有效的测量系统评估和比较将提高我们做出关于使用哪些测量系统和如何改进现有测量系统的明智决定的能力。