R语言关于回归系数的解释

原创

拓端

修改于 2020-08-24 02:04:10

91200

代码可运行

文章被收录于专栏：拓端tecdat拓端tecdat

运行总次数：0

代码可运行

原文链接：http://tecdat.cn/?p=10076

除非我们打算提出因果主张，否则我们应该像描述虚拟变量那样解释连续变量的回归系数。

一条有用建议是，以预测的方式解释回归系数。要了解它们的含义，让我们考虑一个示例。

预测学生表现

hsb <- read.csv("datasets/hsb_comb_full.csv")names(hsb)[1] "schoolid" "minority" "female"   "ses"      "mathach"  "size"     "sector"   [8] "pracad"   "disclim"  "himinty"  "MEANSES"  "N_BREAK"  "sesdev"   "myschool"# Let's go with the first school, and the first 5 student-level variableshsb <- hsb[hsb$schoolid == hsb$schoolid[1], 1:5]summary(hsb)schoolid       minority           female            ses             mathach      Min.   :1224   Min.   :0.00000   Min.   :0.0000   Min.   :-1.6580   Min.   :-2.832  1st Qu.:1224   1st Qu.:0.00000   1st Qu.:0.0000   1st Qu.:-0.8830   1st Qu.: 3.450  Median :1224   Median :0.00000   Median :1.0000   Median :-0.4680   Median : 8.296  Mean   :1224   Mean   :0.08511   Mean   :0.5957   Mean   :-0.4344   Mean   : 9.715  3rd Qu.:1224   3rd Qu.:0.00000   3rd Qu.:1.0000   3rd Qu.:-0.0330   3rd Qu.:16.370  Max.   :1224   Max.   :1.00000   Max.   :1.0000   Max.   : 0.9720   Max.   :23.584  # Mathach, ses and female seem to have some variability# Let's predict math achievement using female (dummy), ses (continuous)lm(mathach ~ female + ses, hsb)Call:lm(formula = mathach ~ female + ses, data = hsb)Coefficients:(Intercept)       female          ses       12.092       -2.062        2.643