r - 在 glm-output 中进行调用 getOption ("width")

Question

在我看来，call 类不尊重 getOption('width')。如果我想在 knitr 中打印模型的 summary，这会使长时间的调用变得丑陋。

有什么办法可以解决这个问题吗？

这是一个小例子:

dataframe <- data.frame(response = seq(10),factor1 = seq(10),
                        factor2 = seq(10), factor3 = seq(10))

model <- glm("response ~ factor1 + factor2 + factor3",
             data = dataframe,
             family = Gamma(link = 'log'))

给出了(丑陋的)输出:

Call:  glm(formula = response ~ factor1 + factor2 + factor3, family = Gamma(link = "log"), 
    data = dataframe)

Coefficients:
(Intercept)      factor1      factor2      factor3  
     0.2923       0.2253           NA           NA  

Degrees of Freedom: 9 Total (i.e. Null);  8 Residual
Null Deviance:      3.886 
Residual Deviance: 0.4238   AIC: 33.05

我发现了类似的问题:Is it possible to make print.formula respect the environment width option? 这样我就能够通过

获取 model$call

strwrap(capture.output(print(model$call)))
## [1] "glm(formula = \"response ~ factor1 + factor2"
## [2] "+ factor3\", family = Gamma(link = \"log\"),"
## [3] "data = dataframe)" 

当使用换行符折叠时，cat 会提供漂亮的打印输出:

cat(paste(
    strwrap(capture.output(print(model$call)))
    ,collapse = "\n"))
## glm(formula = "response ~ factor1 + factor2
## + factor3", family = Gamma(link = "log"),
## data = dataframe)

但我无法将变量分配给cat，即执行类似的操作

model$call <- cat(paste(
    strwrap(capture.output(print(model$call)))
    ,collapse = "\n"))
## glm(formula = "response ~ factor1 + factor2
## + factor3", family = Gamma(link = "log"),
## data = dataframe)
model$call
## NULL

Answer 1

您可以再次使用 capture.output将结果分配给变量:

xx <- paste(
    strwrap(capture.output(print(model$call)))
    ,collapse = "\n"))

model$call <- capture.output(cat(xx))