BAYSIG: a platform for Bayesian analysis of large and complex datasets

目前的数据分析和统计方法可能不够灵活，浪费信息。贝叶斯统计提供了一个替代框架，通过潜在的随机，物理和生物元素的模型来查看数据。测量、假设检验和决策考虑了所有可用的信息，并且可以证明是最优的，但这些强大的技术在很大程度上仍然无法使用。在我们的BBSRC研究中，我们将贝叶斯方法与专业编程语言相结合，构建了一个独特的数据管理和分析平台（BAYSIG）。我们将扩展这个平台，并证明它可以为大型复杂的生物和金融数据集提供重要的新见解。我们将建立一个心跳记录的贝叶斯模型，并表明我们可以比目前更精确地识别真实的心跳信号。这些分析和BAYSIG的其他发展将通过用户友好的网络界面进行商业化。

Current approaches to data analysis and statistics can be inflexible and wasteful of information. Bayesian statistics provide an alternative framework that views data through models of the underlying random, physical and biological elements. Measurements, hypothesis tests and decisions take into account all the available information and are provably optimal, but these powerful techniques remain largely inaccessible. In our BBSRC research we combined Bayesian methods with specialist programming languages to build a unique platform (BAYSIG) for data management and analysis. We will extend this platform and prove that it can provide important new insights into large and complex datasets, both biological and financial. We will build a Bayesian model of heart beat recordings and show that we can characterise real heart beat signals with greater precision than is currently possible. These analyses and other developments of BAYSIG will be commercialised through a user-friendly web interface.