在我写文章画图时经常遇到的一个问题是：ggplot2 坐标轴的输入不支持输入数据框的变量名，通常会报错找不到对象

🌰：问题描述

data: early senate poll

library(tidyverse) # general tasks
library(broom) # tidy model output
library(ggthemes) # style the plots

poll_data <- read_csv("https://raw.githubusercontent.com/fivethirtyeight/data/master/early-senate-polls/early-senate-polls.csv")

glimpse(poll_data)
## Observations: 107
## Variables: 4
## $ year                  <int> 2006, 2006, 2006, 2006, 2006, 2006, 2006...
## $ election_result       <int> -39, -10, -9, -16, 40, 10, -2, -41, -31,...
## $ presidential_approval <int> 46, 33, 32, 33, 53, 44, 37, 39, 42, 33, ...
## $ poll_average          <int> -28, -10, -1, -15, 39, 14, 2, -22, -27, ...

background: there is a strong correlation between polling numbers and the ultimate result of an election

构建模型： 线性模型

poll_lm <- lm(election_result ~ poll_average, data = poll_data)

summary(poll_lm)
## 
## Call:
## lm(formula = election_result ~ poll_average, data = poll_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -29.4281  -5.0197   0.5601   6.1364  17.9357 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -0.89110    0.76969  -1.158     0.25    
## poll_average  1.04460    0.03777  27.659   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 7.93 on 105 degrees of freedom
## Multiple R-squared:  0.8793, Adjusted R-squared:  0.8782 
## F-statistic:   765 on 1 and 105 DF,  p-value: < 2.2e-16

写个函数画出因变量和自变量的关系

结果出现了一个令我费解的报错

Error in FUN(X[[i]], …) : object ‘poll_average’ not found

我不断地检查我的拼写，直到我开始怀疑人生

解决办法：Define aesthetic mappings programatically

plot_model <- function(mod, explanatory, response, .fitted = ".fitted") {
  augment(mod) %>%
  ggplot() +
    geom_point(aes_string(x = explanatory, y = response), color = "#2CA58D") +
    geom_line(aes_string(x = explanatory, y = .fitted), color = "#033F63") +
    theme_solarized() +
    theme(axis.title = element_text()) +
    labs(x = "Poll average", y = "Election results")
}

plot_model(poll_lm, "poll_average", "election_result")