如何将因子更改为数字变量或以其他方式处理我在线性回归中遇到的这个错误

Question

尝试使用此数据集 mro.csv 运行线性回归模型，但是当我运行 lm() 时，它给出了错误消息：

1: In model.response(mf, "numeric") :
  using type = "numeric" with a factor response will be ignored
2: In Ops.factor(y, z$residuals) : ‘-’ not meaningful for factors

不确定数据集的哪些部分是因子而不是数字，所有数据都是数字，列名除外。也不确定“-”对因子部分没有意义是什么，因为数据集中也没有 -。

不知道如何共享数据集，但这是谷歌表中的 csv： mro.csv

> raw <- read.csv("/Users/cpt.jack/Downloads/mro.csv",header<-F,sep<-",") 
> colnames(raw)<- c("inlf","hours","kidslt6","kidsge6","age", "educ",  "wage", "repwage",             "hushrs", "husage", "huseduc","huswage",  "faminc",  "mtr",  "motheduc",  "fatheduc",    "unem","city", "exper",  "nwifeinc",  "lwage",  "expersq")  
> 
> 
> dim(raw)
[1] 753  22
> 
> set.seed(88)
> raw  <- raw[sample(nrow(raw)),]
> 
> 
> raw1<-raw[raw$inlf==1,]
> dim(raw)
[1] 753  22
> dim(raw1)
[1] 428  22
> 
> 
> reg1 <- lm(wage~ hours + kidslt6 + kidsge6 + age + educ + hushrs + husage + huseduc + huswage
+mtr+motheduc+fatheduc+unem
+exper+nwifeinc, data=raw1)

Warning messages:

1: In model.response(mf, "numeric") :
  using type = "numeric" with a factor response will be ignored

2: In Ops.factor(y, z$residuals) : ‘-’ not meaningful for factors
> reg1 <- lm(wage~ hours,data=raw1)

Answer 1

wage和lwage被视为factor s，因为它们包含值"." 不能解析为数字。 该值可以手动处理。

df <- read.csv(
  "~/Downloads/mro.csv",
  header = FALSE,
  stringsAsFactors = FALSE,
  col.names = c(
    "inlf", "hours", "kidslt6", "kidsge6", "age", "educ",  "wage",
    "repwage", "hushrs", "husage", "huseduc", "huswage",  "faminc",
    "mtr",  "motheduc",  "fatheduc", "unem", "city", "exper",
    "nwifeinc", "lwage", "expersq"
  )
)

df$wage <- as.numeric(ifelse(df$wage == ".", 0, df$wage))
df$lwage <- as.numeric(ifelse(df$lwage == ".", 0, df$lwage))

现在lm应该可以正常运行了。

df <- df[sample(nrow(df)), ]
df1 <- df[df$inlf == 1, ]

reg1 <- lm(
  wage ~ hours + kidslt6 + kidsge6 + age + educ + hushrs + husage + huseduc +
         huswage + mtr + motheduc + fatheduc + unem + exper + nwifeinc,
  data = df1
)