Scholarlink

<p>This paper introduces an open-source Python package for simplified, customizable computerized adaptive testing (CAT) using Bayesian methods for ability estimation. It addresses the lack of sophisticated packages for CAT in the Python programming language. Moreover, it bridges the gap between the construction and simulation of adaptive tests and their practical application by providing a dedicated API for integration with experiment software. Thereby, it eliminates the need for major code rewrites when transitioning from simulated to real-world adaptive testing. By leveraging Python’s object-oriented programming approach, such as abstract classes, protocols, and inheritance, the package allows for easy extension and customization of its functionality. For example, Bayesian estimators can be modified to incorporate custom priors. This paper outlines the relevance and practical use of the <Emphasis FontCategory="NonProportional">adaptivetesting</Emphasis> package through a walkthrough example. The package is fully documented, and its source code is published on GitHub. It is also available on the Python Package Index (PyPi) and conda-forge thus it can easily be installed using Python’s package manager <Emphasis FontCategory="NonProportional">pip</Emphasis> or <Emphasis FontCategory="NonProportional">conda</Emphasis>. Leveraging R’s <Emphasis FontCategory="NonProportional">reticulate</Emphasis> package, <Emphasis FontCategory="NonProportional">adaptivetesting</Emphasis> can also be accessed from within RStudio.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Customizable Bayesian adaptive testing with Python – The adaptivetesting package

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月24日

查看全文
  • Jonas Engicht,
  • R. Maximilian Bee,
  • Tobias Koch

摘要

This paper introduces an open-source Python package for simplified, customizable computerized adaptive testing (CAT) using Bayesian methods for ability estimation. It addresses the lack of sophisticated packages for CAT in the Python programming language. Moreover, it bridges the gap between the construction and simulation of adaptive tests and their practical application by providing a dedicated API for integration with experiment software. Thereby, it eliminates the need for major code rewrites when transitioning from simulated to real-world adaptive testing. By leveraging Python’s object-oriented programming approach, such as abstract classes, protocols, and inheritance, the package allows for easy extension and customization of its functionality. For example, Bayesian estimators can be modified to incorporate custom priors. This paper outlines the relevance and practical use of the adaptivetesting package through a walkthrough example. The package is fully documented, and its source code is published on GitHub. It is also available on the Python Package Index (PyPi) and conda-forge thus it can easily be installed using Python’s package manager pip or conda. Leveraging R’s reticulate package, adaptivetesting can also be accessed from within RStudio.