24小时热门版块排行榜

>论坛更新日志 (3906)
>文献求助 (226)
>虫友互识 (212)
>基金申请 (168)
>导师招生 (146)
>休闲灌水 (99)
>论文投稿 (74)
>硕博家园 (71)
>教师之家 (50)
>考研 (43)
>考博 (40)
>博后之家 (39)
>公派出国 (35)
>找工作 (34)
>招聘信息布告栏 (31)
>论文道贺祈福 (29)

返回列表

clivia

金虫 (小有名气)

应助: 0 (幼儿园)
金币: 1308
散金: 417
红花: 2
帖子: 168
在线: 141.8小时
虫号: 414288
注册: 2007-06-27
性别: GG
专业: 景观与区域生态学

[求助] R求助

最近用R处理数据，有以下几个问题问高手:
1,You used a linear model. This assumes that your data (the residuals) are normally distributed? If not, have you tried to transform the residuals so that they are normally distributed? If residuals are not normally distributed and cannot be transformed so that they become normally distributed, have you tried a generalized linear model (GLM) with an appropriate error structure (given by the "family"-command in glm - in R: model <- glm (y ~ x), family= "gaussian / binomial / Gamma / etc."

对于统计数据中有缺失的部分，我一律用R代替，请问如何判断这些数值的出现是否为"gaussian / binomial / Gamma ？还是要一个一个尝试？
gaussian / binomial / Gamma 怎么体现在R语句中？具体怎么操作？

2, What y and what x did you use in "lm(formula = y ~ x)"? You can use several explanatory variables (x) in a lm or glm but only one response variable (y)?

3,With the ratio of annual species and perennial species, I suggest that you use the logarithm of this ratio: log (annuals / perennials) - this is usually done with ratio-data
对于较小的数值（0—1之间），一般取log，请问是取什么为底的函数？10，2，还是e?
谢谢！

回复此楼