【PaperReading】The permutation testing approach: a review

这篇综述探讨了排列检验方法的广泛应用和解决复杂多变量问题的能力。与传统参数和非参数方法相比,排列检验在某些条件下提供更精确的非参数性质,并且在处理依赖性数据和多变量问题时特别有用。文章回顾了这种方法的理论和在各种实验及观察情况下的应用,并讨论了其特性,如组合排列检验的功率在样本量和变量数增加时趋于1。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

The permutation testing approach: a review

Fortunato Pesarin, Luigi Salmaso


摘要:

近年来,permutation testing方法在应用数量和解决复杂的多变量问题方面都有所增加。传统的基于参数或基于秩的非参数方法也可以有效地解决大量的测试问题,尽管在相对温和的条件下,它们的permutation对应物通常渐近和最好的排列对应。Permutation tests在条件上下文中基本上具有精确的非参数性质,其中条件化的观察数据作为零假设中的一组足够的统计量进行调节。相反,大多数参数测试的参考零分布仅以渐近方式已知。因此,对于大多数实际感兴趣的样本,排列解决方案可能缺乏效率可以通过缺乏参数对应物的近似来补偿。存在许多复杂的多变量问题(在生物统计学,临床试验,工程,环境,流行病学,实验数据,工业统计学,药理学,心理学,社会科学等中很常见),这些问题在条件框架之外和用于依赖置换测试的非参数组合(NPC)方法。在本文中,我们回顾了这种方法以及在不同实验和观测情况下的许多应用(例如,多边替代,零膨胀数据和随机排序测试),并且我们提出了特定于该方法的属性,例如: 当给定数量的受试者,当变量的数量发

Permutation Entropy (PE) is a measure of complexity or unpredictability in time series data, and it's often used to analyze non-linear dynamics. In MATLAB, you can calculate Permutation Entropy using the `permuteEntropy` function from the `Entropies` toolbox, which comes as part of the Statistics and Machine Learning Toolbox. Here's an overview of how to compute Permutation Entropy in MATLAB: 1. First, ensure that you have the appropriate toolbox installed. If not, install it with `installToolbox('Statistics_Toolbox')`. 2. Load your time series data into a variable, say `data`. This should be a vector or matrix representing your measurements over time. ```matlab data = [your_time_series_data]; ``` 3. Set the parameters for calculating PE, such as embedding dimension (`m`), tolerance for equal values (`r`), and length of the sequence (`lag`). These values depend on the nature of your data and desired analysis. ```matlab embedding_dim = 3; % Example value tolerance = 0.1; % Example value sequence_length = floor(length(data) / lag); % Choose a suitable lag ``` 4. Now, use the `permuteEntropy` function: ```matlab pe = permuteEntropy(data, embedding_dim, tolerance, sequence_length); ``` 5. The output `pe` will be the estimated permutation entropy of your time series. Here's the complete code snippet: ```matlab % Ensure necessary toolbox is installed if ~isToolboxInstalled('Statistics_Toolbox') error('Please install the Statistics and Machine Learning Toolbox.'); end % Load your data data = [your_time_series_data]; % Define parameters embedding_dim = 3; tolerance = 0.1; lag = floor(length(data) / sequence_length); % Assuming no remainder % Calculate permutation entropy sequence_length = lag; pe = permuteEntropy(data, embedding_dim, tolerance, sequence_length); % Display or save the result disp(['Permutation Entropy: ', num2str(pe)]); ```
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值