||
本文为德国慕尼黑科技大学(作者:Kehinde Fawumi)的硕士论文,共117页。
在几乎每个组织的广泛应用中,每天都会产生大量的时间序列数据,然而,这些数据的管理和分析对于那些缺乏编程经验和时间序列分析模型知识的终端用户来说仍然是一个巨大的挑战。尽管有许多用于时间序列分析的工具,但对这些工具的回顾表明,它们通常是为数据专家和分析人员设计的。在本研究中,一个基于web的交互式时间序列软件是为方便最终用户而设计的。软件设计符合面向终端用户的时间序列管理、分析和转换软件的典型特性,本文对时间序列的研究现状及其在私人和公共电子表格中的共性进行了综述,对真实世界电子表格中的时间序列进行了识别,研究结果表明,在EUSES和Enrons语料库中,有14%的电子表格是时间序列。然后,对现有的一些时间序列工具进行了回顾。这篇评论揭示了:1.目前的这些工具中很少有易于最终用户使用的。迄今为止,大多数时间序列工具都是为专业分析人员和数据科学家开发的。2.大多数时间序列工具对时间序列的转换支持较差;这涉及到将时间戳数据转换为时间序列或将时间序列从一个时频级转换为另一个时频级,然后通过对现有工具的回顾,讨论了软件设计的实际功能需求,这些需求构成了本研究软件设计的基础。设计的时间序列软件的使用场景用实物模型来说明,清楚地说明了用户如何使用该软件来有效地管理、转换和分析时间序列数据。
Every day, time series data are generatedin largevolumesin a wide range of applications in nearly every organization.However, the management and analysis of these data still pose a great challenge to end users who have little programming experience andlittle knowledge of time series analysis models. Although, many tools exist for time series analysis, a review of these tools shows that they are usually designed for data experts andanalysts.In this research, an interactive web based time series software isdesigned for ease of use by end users. The software design aligns with typical properties of an end user oriented software for managing, analyzing and transforming time series.Firstly, thisthesis reports on the current state of research on time series and theircommonness in private and public spreadsheets.Time series are identified in real world spreadsheets and results show that 14 percentof spreadsheets in the EUSES corpusand Enrons corpus are time series. Then,a review of some existing time series tools is made. This review reveals that:1.Only a few ofthese tools are easy to use for end users. Hitherto, most time series toolshave been developed for usage by professional analysts and data scientists.2.Most of the tools give poor support for the transformation of time series; which involves the reduction of time-stamped data to time series or the conversion of time series from one level of time frequencyto another.A set of functional requirements for the thesis software is thengenerated from the review of the existing tools. These requirements form the basis for the software design in this research. Theusage scenarios of the time seriessoftware designed areillustrated using mockups.This clearly illustrates how users can work with the software toeffectively manage, transform and analyzetimeseries data.
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-9-1 09:20
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社