如何将变量的正方形作为i（x^2）与循环添加在线性模型中？

Question

我生成了获取数据的东西，这些数据将所有列名添加到线性回归中，添加了它们的交互和该变量的第二度，并使用步骤函数来找到最佳的线性回归模型。我添加了所有列的名称和交互，但是在这里我有一个问题，添加了第二度变量。我使用上述代码，但

Error in (colnames(dpt)[i])^2 : non-numeric argument to binary operator" error

有人可以帮忙吗？我真的很感激。

for (i in colnames(dpt)) {
  lmod <- lmod %>% update(paste('. ~ . + I((colnames(dpt)[i])^2)'),
                          data = dpt)
}

Answer 1

使用sprintf。

fit <- lm(mpg ~ cyl, mtcars)

v <- names(mtcars)[-(1:2)]

res <- lapply(v, \(x) update(fit, sprintf('. ~ . + I(%s^2)', x)))

或者，如果您进入管道：

res <- lapply(v, \(x) fit |> update(sprintf('. ~ . + I(%s^2)', x)))

或者如果没有循环的

res <- vector(mode='list', length=length(v))
for (i in seq_along(v)) {
  res[[i]] <- fit |> update(sprintf('. ~ . + I(%s^2)', v[i]))
}

res
# [[1]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(disp^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl    I(disp^2)  
# 3.543e+01   -2.246e+00   -2.115e-05  
# 
# 
# [[2]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(hp^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl      I(hp^2)  
# 3.705e+01   -2.641e+00   -2.374e-05  
# 
# 
# [[3]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(drat^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl    I(drat^2)  
# 31.6655      -2.4568       0.2745  
# 
# 
# [[4]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(wt^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl      I(wt^2)  
# 35.8818      -1.9282      -0.3423  
# 
# 
# [[5]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(qsec^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl    I(qsec^2)  
# 43.73138     -3.15264     -0.01285  
# 
# 
# [[6]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(vs^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl      I(vs^2)  
# 39.6250      -3.0907      -0.9391  
# 
# 
# [[7]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(am^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl      I(am^2)  
# 34.522       -2.501        2.567  
# 
# 
# [[8]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(gear^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl    I(gear^2)  
# 36.03695     -2.76010      0.08013  
# 
# 
# [[9]]
# 
# Call:
#   lm(formula = mpg ~ cyl + I(carb^2), data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl    I(carb^2)  
# 37.55404     -2.74723     -0.04454