在Stata中，”factor”是否与R中的”princomp”执行相同的功能？

huangapple
go评论114阅读模式
英文:

Does factor in Stata do the same thing as princomp in R?

问题

我需要在R中重新创建一个使用Stata的factor命令创建的变量。

我正在寻找R中类似于Stata的factor命令的相应命令，例如：

  1. factor v1 v2 v3 v4 v5 
  2. score newvar1

根据这份Stata文档，这是一个主成分因子分析。

我尝试使用R中的因子分析 fa() 命令来执行相同的操作。然而，我在语法上遇到了困难。我认为相似的R代码应该如下所示：

  1. library(psych)    
  2. df$newvar1 <- fa(df, values = c(v1, v2, v3, v4, v5))

然而，当我使用mtcars虚拟数据进行测试时，我没有得到一个因子变量作为结果：

  1. mtcars <- mtcars
  2. library('psych')
  3. mtcars$newvar1 <- fa(mtcars, values=c(mpg, cyl, hp))

错误信息：

Error in $<-.data.frame(*tmp*, fa, value = list(residual =
c(0.124759140589603, : replacement has 49 rows, data has 32 In
addition: Warning message: In fa.stats(r = r, f = f, phi = phi, n.obs
= n.obs, np.obs = np.obs, : The estimated weights for the factor scores are probably incorrect. Try a different factor score
estimation method.

我是不是漏掉了某些参数，还是我误解了语法？

英文:

Edit: for spelling and based on initial response, because I am still struggling with the syntax

I need to recreate a variable in R that was created in Stata using the factor command.

I am looking for the analogous command in R to Stata's factor command, written for example as:

  1. factor v1 v2 v3 v4 v5 
  2. score newvar1

which, according to this Stata documentation, is a principal-factor analysis.

I am trying to use the factor analysis fa() command in R do the same thing. However, I am struggling with the syntax. I think the analogous R code to the Stata code above should look something like this:

  1. library(psych)    
  2. df$newvar1 &lt;- fa(df, values = c(v1, v2, v3, v4, v5))

However, when I trial it out using the mtcars dummy data, I do not get a factor variable as a result:

  1. mtcars &lt;- mtcars
  2. library(&#39;psych&#39;)
  3. mtcars$newvar1 &lt;- fa(mtcars, values=c(mpg, cyl, hp))

> Error in $&lt;-.data.frame(*tmp*, fa, value = list(residual =
> c(0.124759140589603, : replacement has 49 rows, data has 32 In
> addition: Warning message: In fa.stats(r = r, f = f, phi = phi, n.obs
> = n.obs, np.obs = np.obs, : The estimated weights for the factor scores are probably incorrect. Try a different factor score
> estimation method.

Am I missing some argument, or am I misinterpreting the syntax?

答案1

得分: 1

以下是R代码的翻译部分：

  1. library(psych)
  2. set.seed(123)
  3. data(mtcars)
  4. f <- fa(mtcars[,c("mpg", "disp", "hp", "qsec")], nfactors=1, SMC=TRUE, fm="pa", rotate="none", max.iter=1)
  5. #> 最大迭代次数已超出
  6. loadings(f)
  7. #> 
  8. #> 因子载荷：
  9. #>      PA1   
  10. #> mpg  -0.860
  11. #> disp  0.872
  12. #> hp    0.928
  13. #> qsec -0.633
  14. #> 
  15. #>                 PA1
  16. #> 因子SS载荷    2.76
  17. #> 方差比例 0.69

以下是Stata代码的翻译部分：

  1. . factor mpg disp hp qsec, pf factors(1)
  2. (obs=32)
  4. 因子分析/相关性                      观测数 = 32
  5.     方法: 主要因子                        保留的因子 = 1
  6.     旋转: (未旋转)                        参数数 = 4
  8.     --------------------------------------------------------------------------
  9.          因子  |   特征值   差异        比例   累积
  10.     -------------+------------------------------------------------------------
  11.         因子1  |      2.75999      2.42226            0.9543       0.9543
  12.         因子2  |      0.33773      0.42422            0.1168       1.0711
  13.         因子3  |     -0.08649      0.03268           -0.0299       1.0412
  14.         因子4  |     -0.11917            .           -0.0412       1.0000
  15.     --------------------------------------------------------------------------
  16.     LR检验: 独立 vs. 饱和:  chi2(6)  =   94.62 Prob>chi2 = 0.0000
  18. 因子载荷 (模式矩阵) 和唯一方差
  20.     ---------------------------------------
  21.         变量 |  因子1 |   唯一性 
  22.     -------------+----------+--------------
  23.              mpg |  -0.8595 |      0.2613  
  24.             disp |   0.8719 |      0.2397  
  25.               hp |   0.9277 |      0.1393  
  26.             qsec |  -0.6327 |      0.5997  
  27.     ---------------------------------------

请注意，fa()函数的fm="pa"选项默认执行迭代主轴因子分析（等同于Stata的ipf选项）。要使fa()函数执行一次性主轴因子分析解决方案，需要设置max.iter=1。

英文:

Here is the equivalent in R:

  1. library(psych)
  2. set.seed(123)
  3. data(mtcars)
  4. f &lt;- fa(mtcars[,c(&quot;mpg&quot;, &quot;disp&quot;, &quot;hp&quot;, &quot;qsec&quot;)], nfactors=1, SMC=TRUE, fm=&quot;pa&quot;, rotate=&quot;none&quot;, max.iter=1)
  5. #&gt; maximum iteration exceeded
  6. loadings(f)
  7. #&gt; 
  8. #&gt; Loadings:
  9. #&gt;      PA1   
  10. #&gt; mpg  -0.860
  11. #&gt; disp  0.872
  12. #&gt; hp    0.928
  13. #&gt; qsec -0.633
  14. #&gt; 
  15. #&gt;                 PA1
  16. #&gt; SS loadings    2.76
  17. #&gt; Proportion Var 0.69

<sup>Created on 2023-01-10 by the reprex package (v2.0.1)</sup>

Stata:

  1. . factor mpg disp hp qsec, pf factors(1)
  2. (obs=32)
  4. Factor analysis/correlation                      Number of obs    =         32
  5.     Method: principal factors                    Retained factors =          1
  6.     Rotation: (unrotated)                        Number of params =          4
  8.     --------------------------------------------------------------------------
  9.          Factor  |   Eigenvalue   Difference        Proportion   Cumulative
  10.     -------------+------------------------------------------------------------
  11.         Factor1  |      2.75999      2.42226            0.9543       0.9543
  12.         Factor2  |      0.33773      0.42422            0.1168       1.0711
  13.         Factor3  |     -0.08649      0.03268           -0.0299       1.0412
  14.         Factor4  |     -0.11917            .           -0.0412       1.0000
  15.     --------------------------------------------------------------------------
  16.     LR test: independent vs. saturated:  chi2(6)  =   94.62 Prob&gt;chi2 = 0.0000
  18. Factor loadings (pattern matrix) and unique variances
  20.     ---------------------------------------
  21.         Variable |  Factor1 |   Uniqueness 
  22.     -------------+----------+--------------
  23.              mpg |  -0.8595 |      0.2613  
  24.             disp |   0.8719 |      0.2397  
  25.               hp |   0.9277 |      0.1393  
  26.             qsec |  -0.6327 |      0.5997  
  27.     ---------------------------------------

Note, the fa() function with fm=&quot;pa&quot; does iterated principal axis factoring by default (equivalent to Stata's, ipf option). To have fa() do the one-shot principal axis factoring solution, set max.iter=1.

huangapple
  • 本文由 发表于 2023年1月9日 03:43:33
  • 转载请务必保留本文链接：https://go.coder-hub.com/75050758.html
  • r
  • stata
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定