列名有特殊字符时的循环回归

Question

如下所示的虚拟数据。我尝试修改循环回归代码以考虑列名称中的空格，但它仍然给出错误。此外，我也不知道如何调整

.x

来解释自变量列名称中的空格。 该代码的目的是一次将第一列与随后的每一列进行回归。即

Canada Price ~ Global Price

，然后

Canada Price ~ Canada Bond Price

等等。

lm.test <- map_dfr(
  set_names(names(df)),
  ~ tidy(lm(as.formula(paste("`",colnames(df[1]),"`","~", .x)),
            data = df)),
  .id = "predictor") 

加拿大价格	全球价格	加拿大2年期债券价格	美国 - 加拿大通胀	油价
-0.030661468	-0.000207995	0.084	-0.075	0.002116
-0.034269304	-0.140212141	0.363	-1.062	-0.06122
0.02371464	0.019987	-0.781	0.273	0.209895
-0.007971484	0.090597341	-0.221	0.068	-0.10657
0.005342716	0.060627149	0.151	0.234	-0.22191
-0.002210376	0.046010712	-0.106	0.13	0.073975
0.018338299	0.019799534	0.036	0.005	0.058091
0.000662471	-0.025356047	0.523	0.045	-0.03765
0.002124411	0.045979342	-0.433	0.282	0.365933
0.034191555	0.043211347	-0.08	0.147	0.113365
-0.011470069	-0.033773176	0.396	-0.037	-0.09753
0.007039847	0.050327089	-0.052	0.151	0.145487
-0.028872959	-0.004501371	0.324	-0.214	0.064282
0.009244373	-0.00092413	0.166	-0.066	0.076961
0.017448201	-0.010638258	-0.179	0.053	0.108548
-0.002379819	0.050111674	0.352	-0.161	-0.11261
-0.002103406	0.03094893	0.121	0.106	0.130575
0.019154969	0.08307375	0.051	0.149	0.041074
-0.00124318	-0.054526954	0.345	0.032	0.079687
-0.001241636	0.003319001	-0.289	0.219	0.100941

---

编辑

数据采用

dput

格式。

df1 <-
  structure(list(
    `Canada Price` = c(-0.030661468, -0.034269304, 0.02371464, -0.007971484, 
                       0.005342716, -0.002210376, 0.018338299, 0.000662471, 0.002124411, 
                       0.034191555, -0.011470069, 0.007039847, -0.028872959, 0.009244373, 
                       0.017448201, -0.002379819, -0.002103406, 0.019154969, -0.00124318, -0.001241636), 
    `Global Price` = c(-0.000207995, -0.140212141, 0.019987, 0.090597341, 
                       0.060627149, 0.046010712, 0.019799534, -0.025356047, 0.045979342,
                       0.043211347, -0.033773176, 0.050327089, -0.004501371, -0.00092413, 
                       -0.010638258, 0.050111674, 0.03094893, 0.08307375, -0.054526954, 0.003319001), 
    `Canada 2Y Bond Price` = c(0.084, 0.363, -0.781, -0.221, 0.151, -0.106, 
                               0.036, 0.523, -0.433, -0.08, 0.396, -0.052, 0.324, 0.166, -0.179, 0.352,
                               0.121, 0.051, 0.345, -0.289), 
    `US - Canada Inflation` = c(-0.075, -1.062, 0.273, 0.068, 
                                0.234, 0.13, 0.005, 0.045, 0.282, 0.147, -0.037, 0.151, -0.214, 
                                -0.066, 0.053, -0.161, 0.106, 0.149, 0.032, 0.219), 
    `Oil Price` = c(0.002116, -0.06122, 0.209895, -0.10657, -0.22191, 0.073975, 
                    0.058091, -0.03765, 0.365933, 0.113365, -0.09753, 0.145487, 0.064282, 
                    0.076961, 0.108548, -0.11261, 0.130575, 0.041074, 0.079687, 0.100941)), 
    class = "data.frame", row.names = c(NA, -20L))

Answer 1

这是对除第一列之外的所有其他列运行回归的解决方案。
该公式是用

reformulate

创建的。

suppressPackageStartupMessages({
  library(magrittr)
  library(purrr)
})

# create the formula with 'value' as regressor,
# it will be the column name after pivoting to long format
fmla <- reformulate("value", response = names(df1)[1L] |> as.name())
# run the regressions
lm_fit_list <- df1 %>%
  tidyr::pivot_longer(-1L) %>%
  split(.$name) %>%
  map(\(df) lm(fmla, data = df))
  
# this gives some statistics
lm_smry_list <- lm_fit_list %>% map(summary)
lm_smry_list %>% map(coef)
#> $`Canada 2Y Bond Price`
#>                 Estimate  Std. Error    t value   Pr(>|t|)
#> (Intercept)  0.001736073 0.003667346  0.4733866 0.64162781
#> value       -0.025789455 0.011690588 -2.2060015 0.04061964
#> 
#> $`Global Price`
#>                 Estimate  Std. Error    t value   Pr(>|t|)
#> (Intercept) -0.001515084 0.003712875 -0.4080622 0.68804100
#> value        0.164831106 0.070018144  2.3541199 0.03012634
#> 
#> $`Oil Price`
#>                 Estimate  Std. Error    t value  Pr(>|t|)
#> (Intercept) -0.001424249 0.004112478 -0.3463238 0.7331147
#> value        0.046411939 0.030404220  1.5264966 0.1442663
#> 
#> $`US - Canada Inflation`
#>                 Estimate  Std. Error    t value    Pr(>|t|)
#> (Intercept) 0.0001923356 0.003176245 0.06055441 0.952381362
#> value       0.0393945114 0.011325922 3.47826091 0.002683133

^{创建于 2024-02-23，使用 reprex v2.0.2}