代码之家 › 专栏 › 技术社区 › Francisco

关于错误的问题:$operator对原子向量无效

package function r

0

Francisco · 技术社区 · 2 年前

我创建了这个函数来自动进行多项回归:

multinom<-function(form,ref,db){
  library(VGAM)
  mod<-vglm(formula(form),multinomial(refLevel = ref),data = db)
  t1<-exp(cbind(OR = coef(mod), confint(mod)))
  p_val<-coef(summary(mod))[,4]
  x<-data.frame(cbind(t1,p_val))
  x$prob<-x[,1]/(1+x[,1])
  colnames(x)<-c("OR","CI_min","CI_max","P_val","Probablity")
  return(x)
}

当这个函数在脚本中使用时,它运行得很好。我还在正在编写的程序包中添加了此功能:

##############################
#' Statistical analysis: Multinomial Regression
#'
#'The purpose of this function is to calculate the Odds Ratio for a statistical analysis that involves a categorical dependent variable, with more than two unordered levels or classes, and different independent variables of any kind.
#'The Odds Ratio is obtained using a multinominal logit regression. This is an extension of the binomial logistic regression to allow for a categorical dependent variable with more than two unordered categories.
#'In multinomial regression we need to define a reference variable category. The model will determine several binomial distribution parameters with respect to the reference category.
#' @param form model formula. For example: `dep_var~conf_var1+conf_var2+conf_var3`.
#' @param ref reference category from the dependent variable. If its a character variable the reference level must be included in `""`.
#' @param df D.B_Final database obtained from the data incorporation step.
#' @return A table with the Odds ratio, its 95% CI and the associated p-value.
#' @export
#' @examples
#' # Basic Usage of Multinomial Regression.
#' multinom(cov_2~cov_3+var_con,"A",D.B_Final)
#'
multinom<-function(form,ref,db){
  library(VGAM)
  mod<-vglm(formula(form),multinomial(refLevel = ref),data = db)
  t1<-exp(cbind(OR = coef(mod), confint(mod)))
  p_val<-coef(summary(mod))[,4]
  x<-data.frame(cbind(t1,p_val))
  x$prob<-x[,1]/(1+x[,1])
  colnames(x)<-c("OR","CI_min","CI_max","P_val","Probablity")
  return(x)
}

当此功能在相应的包中使用时,具体为 zoombedo::multinorm ,我遇到错误消息: Error in object$coefficients: $ operator is invalid for atomic vectors.

有人能帮我了解一下这里发生了什么吗?

1 回复 | 直到 2 年前

1

4

Ben Bolker 2 年前

我没有一个完美的解决方案,但我想我可以解释这个问题。几乎可以肯定的是,这个错误来自于 coef(summary(mod))[,4] 这应该正在呼叫 coef() 类的对象的方法 summary.vglm :

selectMethod("coef", "summary.vglm")
Method Definition:

function (object, ...) 
object@coef3
<bytecode: 0x5617c6d9d8e8>
<environment: namespace:VGAM>

这显然在包上下文中找不到。相反,你得到的是 stats:::coef.default() :

function (object, complete = TRUE, ...) 
{
    cf <- object$coefficients
    if (complete) 
        cf
    else cf[!is.na(cf)]
}

由于对象没有 $coefficient 元素,则会出现错误。

解决此问题的方法是访问 @coef3 直接插入插槽。

使用 importMethodsFrom("VGAM", "summary") 在您的 NAMESPACE 文件(或 @importMethodsFrom 如果你使用的是roxygen),也许 importClassesFrom("VGAM", "summary.vglm") 可以解决问题(请参阅 R Extensions 文档),但我还没有测试。