An organisation wishing to conduct data analytics to support day-to-day decision making often needs a system to help analysts represent and maintain knowledge about research variables, datasets or analytical models, and effectively determine the best combination to use when solving the problem at hand. Often, such knowledge is not explicitly captured by the organisation. To address this problem, this paper presents the design of an innovative Information Technology (IT) platform which enables data sharing between different analytics models and provides the ability to extend or customise models or data sources without necessarily involving the analysts who created them. It can make analytics knowledge readily available and modifiable for future use and problem-solving by analysts and other stakeholders. In the context of our work, we organise analytics knowledge around the concept of a research variable, which analysts often use when defining and proving a hypothesis. By focusing on such a concept, this platform is particularly suited to develop empirical data analytics applications in any domain. This paper presents the architecture of this platform, including the knowledge base and the Application Programming Interface (API) layer. Capabilities of this platform are illustrated through a software prototype and a use case on property price prediction across Sydney, Australia.

希望进行数据分析以支持日常决策的组织通常需要一个系统来帮助分析师表示和维护有关研究变量，数据集或分析模型的知识，并有效地确定解决当前问题时要使用的最佳组合。通常，组织不会明确地获取此类知识。为了解决这个问题，本文介绍了一种创新的信息技术（IT）平台的设计，该平台能够在不同分析模型之间共享数据，并具有扩展或自定义模型或数据源的能力，而不必涉及创建它们的分析师。它可以使分析知识易于获得和修改，以供分析人员和其他利益相关者将来使用和解决问题。在我们的工作中，研究变量，分析人员在定义和证明假设时经常使用。通过关注这一概念，该平台特别适合在任何领域中开发经验数据分析应用程序。本文介绍了该平台的体系结构，包括知识库和应用程序编程接口（API）层。该平台的功能通过软件原型和澳大利亚悉尼房价预测的用例进行了说明。