This paper presents a high-level description of the R software package mixdistreg to fit mixture of experts distributional regression models. The proposed framework is implemented in R using the deepregression software template, which is based on TensorFlow and follows the neural structured additive learning principle. The software comprises various approaches as special cases, including mixture density networks and mixture regression approaches. Various code examples are given to demonstrate the package's functionality.
deepregression: a Flexible Neural Network Framework for Semi-Structured Deep Distributional Regression
LMU Munich(慕尼黑大学);Helmholtz AI(亥姆霍兹人工智能研究所);ETH Zurich(苏黎世联邦理工学院);University of Zurich(苏黎世大学);Humboldt-Universität zu Berlin(柏林洪堡大学);Helmholtz Zentrum Munich(慕尼黑亥姆霍兹中心);Flatiron Institute(弗莱提荣研究所)
Federal University of São Carlos, University of São Paulo(巴西圣保罗联邦大学、圣保罗大学);Institute of Mathematics and Computer Sciences, University of São Paulo(圣保罗大学数学与计算机科学学院)
Generating Synthetic Electronic Health Record Data: a Methodological Scoping Review with Benchmarking on Phenotype Data and Open-Source Software
Department of Biostatistics, University of Michigan(生物统计学系,密歇根大学);Department of Biomedical Informatics and Data Science, Yale University(生物医学信息学与数据科学系,耶鲁大学);Department of Biostatistics, Yale University(生物统计学系,耶鲁大学)
我们提出了一种新的框架,用于拟合添加剂分位数回归模型,它提供了良好的校准推断的条件分位数和快速自动估计的平滑参数,模型结构不同的分布GAM,同时保持相同的数值效率和稳定性。所提出的方法在统计上是严格的,在计算上是有效的,因为它们是基于Bissiri et al.(2016)的一般信念更新框架,以基于损失的推理,但通过调整Wood et al.(2016)的稳定拟合方法来计算。我们展示了弹球损失是如何统计次优相对于一个新的平滑概括,这也提供了快速估计方法。此外,我们提供了一种新的校准方法,有效地选择“学习率”平衡损失与平滑先验在推理过程中,从而获得可靠的分位数不确定性估计。我们的工作是出于概率电力负荷预测的应用程序,在这里使用,以证明所提出的方法。这里描述的方法由qgam R包实现,该包可在Comprehensive R Archive Network(CRAN)上获得。