zhang2sha的个人博客分享 http://blog.sciencenet.cn/u/zhang2sha

博文

他们重新分析了PlosONE上一篇论文的数据,转手发了Nature!

已有 580 次阅读 2024-3-5 19:13 |系统分类:科研笔记

把发表在plosONE上的一篇论文的数据重新分析一遍,然后发了Nature, 这竟然不是琼瑶片。下面这篇去年的nature文章就是如此!看来搞研究,格局一定要打开啊,再也不能说PlosONE发的都是烂文章了(我自己其实也发过哈哈)。

图1.jpg

这篇nature文章的摘要里说“Here, we reanalysed 27 years of insect biomass data from Hallmann et al.1, using sample-specific information on weather conditions during sampling and weather anomalies during the insect life cycle.”

而Hallmann et al的这篇文章,就是2017年发表在曾经红极一时的著名期刊PlosONE上的一篇论文。

图2.png

没错,这篇nature的数据完全来源于这篇PlosONE论文,只不过是更新了分辨率更高的气象数据,

To obtain consistent data for training and validation data in time and space, we did not use the meteorological data from the original work, but extracted updated data from E-OBS v.25.0e with a 0.1° regular grid for each individual sample.

 

然后作者采用新的模型分析了各种气象因子对飞虫生物量的影响,就形成了这篇Nature论文。不得不说,实在是高,搞科研,还是要破除任何固有关键,以回答科学问题为核心啊!

 

回到数据分析,这篇Nature通篇采用了log-linear Gaussian additive models以及log-linear Gaussian additive mixed models,也是非线性模型应用的具体案例。数据分析中还涉及到了非线性模型的offset设置,模型对比,预测、结果制图、制表、Validation等,值得我们学习,参考。

 

比如其中用到的log-linear Gaussian additive models,模型结构为:

 

hallmann_plus <- gam(biomass ~ s(meandaynr) + offset(log(todaynr - fromdaynr)) + s(E, N, bs = "tp") +

                               nHerbs + nTrees + Light + ellenTemperature +

                               Arableland + Forest + Grassland + Water +

                               Tmean_c * Psum_c +

                               Tmean_anomaly_april_current * Psum_anomaly_april_current +

                               Tmean_anomaly_april_prev * Psum_anomaly_april_prev +

                               Tmean_anomaly_winter * Psum_anomaly_winter +

                               Tmean_anomaly_meandaynr_prev * Psum_anomaly_meandaynr_prev,

                      family = gaussian(link = "log"),

                      method = METHOD,

                      data = training)

 

 

文中用到的log-linear Gaussian additive mixed models,模型结构为:

hallmann_plus_RE <- gam(biomass ~ s(meandaynr) + offset(log(todaynr - fromdaynr)) + s(E, N, bs = "tp") +

                               s(fyear, bs = "re") +  ### add temporal random effect

                               nHerbs + nTrees + Light + ellenTemperature +

                               Arableland + Forest + Grassland + Water +

                               Tmean_c * Psum_c +

                               Tmean_anomaly_april_current * Psum_anomaly_april_current +

                               Tmean_anomaly_april_prev * Psum_anomaly_april_prev +

                               Tmean_anomaly_winter * Psum_anomaly_winter +

                               Tmean_anomaly_meandaynr_prev * Psum_anomaly_meandaynr_prev,

                      family = gaussian(link = "log"),

                      method = METHOD,

                      data = training)

 

其中s(fyear, bs = "re")即为随机效应部分的设置。

欢迎大家关注我的个人微信公众号“二傻统计”

二傻统计.jpg



https://blog.sciencenet.cn/blog-3442043-1424197.html

上一篇:Meta分析,再登Science!
下一篇:混合模型为何能如此风靡?
收藏 IP: 111.197.238.*| 热度|

1 刘全慧

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2024-5-28 21:39

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部