创建一个for循环，其中我按R中的列标识数据 [英] Creating a for-loop where i identifies data by column in R

问题描述

我正在尝试使用一个for循环来创建一系列GAM模型的摘要，这些模型使用一年中的平滑术语(DOY)和一个结合了天气的线性预测变量(有其中有很多).我做了一个运行和输出模型摘要的函数:

  gamlin<-function(x){m<-gam(Log10E〜s(DOY)+ x，data = eggseasongam)返回(摘要(m))} 

我认为我想在一个for循环中利用以上内容，该循环将第4到173列中的天气预报作为x，但是要努力做到这一点.任何建议将不胜感激.

谢谢，迈克

解决方案

我头顶上的2种方法可能是:

  t< -data.frame(c1 = rnorm(10)，c2 = rnorm(10)，c3 = rnorm(10)) 

1-使用 formula ，它允许您将字符串作为公式提供

 >f< -function(d，x)lm(公式(paste("c1"，x，sep =〜"))，data = d)>f(t，"c2")称呼:lm(公式=公式(paste("c1"，x，sep =〜"))，数据= d)系数:(截取)c2-0.1567 -0.4654 

2-或者直接在列中输入

 >f2 ＜函数(d，x)lm(c1〜d [，x]，data ＝ d)>f2(t，"c2")称呼:lm(公式= c1〜d [，x]，数据= d)系数:(截取)d [，x]-0.1567 -0.4654 

一旦确定了动态回归函数的方法，就可以将其置于for循环中

  for(col in 4:173)f(t，col) 

可能是，您想以某种方式使用结果，而不是仅打印摘要(也许取系数并将其放入数据帧中并对其进行图形绘制，或沿这些方式绘制图形.在这种情况下，我建议签出 plyr 和 reshape 包，具体来说，每个包中的函数 melt 和 ddply

在这种情况下，您可以执行以下操作

 库(融化)图书馆(plyr)yvar< -t [，1]xvars< -melt(t [，-1])>头(xvars)可变值1立方厘米-0.82002632立方厘米-1.53​​592203立方厘米-0.21079134 c2 -0.29502635 c2 0.82319896 c2 -0.5971358betas <-ddply(xvars，.(variable)，summary，beta = coefficients(lm(yvar〜value))[2]) 

I am trying to use a for-loop to create a series of summaries of GAM models of variation in capture rate of an invertebrate species using a smoothed term for day of year (DOY), and a linear predictor incorporating weather (there are many of these). I have made a function to run and output a model summary:

gamlin <- function(x) {
  m <- gam(Log10E ~ s(DOY) + x, data=eggseasongam)
  return(summary(m))
}

I think that I want to utilize the above in a for-loop that will sequentially take as x the weather predictors in columns 4 through 173, but am struggling to do this. Any suggestions would be very much appreciated.

Thanks, Mike

解决方案

2 approaches off the top of my head could be:

t<-data.frame(c1=rnorm(10),c2=rnorm(10), c3=rnorm(10))

1- use formula, which allows you to provide a character string as a formula

> f<-function(d,x) lm(formula(paste("c1",x,sep="~")), data=d)
> f(t,"c2")

Call:
lm(formula = formula(paste("c1", x, sep = "~")), data = d)

Coefficients:
(Intercept)           c2
    -0.1567      -0.4654

2- Alternatively feed in the column directly

> f2<-function(d,x) lm(c1 ~ d[,x], data=d)
> f2(t,"c2")

Call:
lm(formula = c1 ~ d[, x], data = d)

Coefficients:
(Intercept)       d[, x]
    -0.1567      -0.4654

Once you have settled on an approach for a dynamic regression function you can put it in a for-loop

 for(col in 4:173) f(t,col)

Chances are, you want to use the results somehow, rather than just print a summary (Perhaps take the coefficients and put them in a dataframe and graph them, or something along those lines. In which case I suggest checking out plyr and reshape packages and specficially the functions melt and ddply within each

In this case you could do as follows

library(melt)
library(plyr)

yvar<-t[,1]
xvars<-melt(t[,-1])
> head(xvars)
  variable      value
1       c2 -0.8200263
2       c2 -1.5359220
3       c2 -0.2107913
4       c2 -0.2950263
5       c2  0.8231989
6       c2 -0.5971358

betas<-ddply(xvars, .(variable), summarize, beta=coefficients(lm(yvar ~ value))[2])

这篇关于创建一个for循环，其中我按R中的列标识数据的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！