SBIR PHASE II: Statistical Analysis and Software for Long Memory Processes

*9801614布鲁斯这个小型企业创新研究第二阶段项目将对长记忆过程的统计方法进行新的创新研究。这些过程可用于对地球物理、海洋学、天文学、经济学、生理学和其他学科中常见的时间序列进行建模。长记忆过程的本质是，随着时间间隔的增加，观测之间的相关性缓慢降低。对这些使用标准时间序列方法可能会导致不正确的统计分析；例如，平均值的可信区间可能小得不切实际。这项研究项目将在建立完整的长记忆过程方法学方面取得重要进展，这是对科学的基础性贡献。最终目标是将这项研究和现有方法纳入一个面向对象的软件工具包，用于建立长记忆时间序列的模型，供广泛的统计学家、数据分析员和科学家使用。如果成功，这项研究将产生一个软件工具包，为建立长记忆时间序列模型提供完整的统计方法。该工具包将以S+模块、MathCAD功能包和电子书的形式销售。此外，还将单独开发和销售实践者案例手册。***

*** 9801614 Bruce This Small Business Innovation Research Phase II project will conduct new and innovative research into statistical methodology for long memory processes. These processes can be used to model time series commonly encountered in geophysics, oceanography, astronomy, economics, physiology and other disciplines. The essence of a long memory process is that correlation between observations decreases slowly as separation in time increases. Use of standard time series methodology with these can lead to incorrect statistical analysis; e.g., confidence intervals for the mean can be unrealistically small. This research project will make important advances toward a complete methodology for long memory processes, a fundamental contribution to science. The ultimate objective is to implement the research and existing methodology into an object-oriented software tool kit for modeling long memory time series for use by a broad range of statisticians, data analysts and scientists. If successful, the research will lead to a software tool kit offering complete statistical methodology for modeling long memory time series. The tool kit will be marketed as an S-PLUS module and as a MathCAD function pack and electronic book. In addition, a practitioner's casebook of examples will be developed and marketed separately. ***