Question

我正在尝试运行plm以查看积极，消极和中立的类别对股票价格的影响。

DATE <- c("1","2","3","4","5","6","7","1","2","3","4","5","6","7")
COMP <- c("A", "A", "A", "A", "A", "A", "A", "B", "B", "B", "B", "B", "B", "B")
RET <- c(-2.0,1.1,3,1.4,-0.2, 0.6, 0.1, -0.21, -1.2, 0.9, 0.3, -0.1,0.3,-0.12)
CLASS <- c("positive", "negative", "neutral", "positive", "positive", "negative", "neutral", "positive", "negative", "negative", "positive", "neutral", "neutral", "neutral")
df <- data.frame(DATE, COMP, RET, CLASS, stringsAsFactors=F)

df

#    DATE COMP   RET    CLASS
# 1     1    A -2.00 positive
# 2     2    A  1.10 negative
# 3     3    A  3.00  neutral
# 4     4    A  1.40 positive
# 5     5    A -0.20 positive
# 6     6    A  0.60 negative
# 7     7    A  0.10  neutral
# 8     1    B -0.21 positive
# 9     2    B -1.20 negative
# 10    3    B  0.90 negative
# 11    4    B  0.30 positive
# 12    5    B -0.10  neutral
# 13    6    B  0.30  neutral
# 14    7    B -0.12  neutral

如果我运行模型，输出只显示两个估计值（中性和正值）。我怎么能看到班级负面的估计？我认为这与傻瓜有关。但是，对于否定阶级，至少应该至少有一条“拦截”线？

mymodel <- plm(RET ~ CLASS, data=df,
              index = c("DATE", "COMP"), 
              model="within", 
              effect="time")

summary(mymodel)

# Oneway (time) effect Within Model

# Call:
# plm(formula = RET ~ CLASS, data = df, effect = "time", model = "within", 
#     index = c("DATE", "COMP"))

# Balanced Panel: n=7, T=2, N=14

# Residuals :
#    Min. 1st Qu.  Median 3rd Qu.    Max. 
# -2.1500 -0.4620 -0.0791  0.7540  1.9300 

# Coefficients :
#               Estimate Std. Error t-value Pr(>|t|)
# CLASSneutral   0.35818    0.81581  0.4390    0.670
# CLASSpositive -0.56418    0.81581 -0.6916    0.505

# Total Sum of Squares:    16.79
# Residual Sum of Squares: 14.694
# R-Squared      :  0.12486 
#       Adj. R-Squared :  0.089183 
# F-statistic: 0.713347 on 2 and 10 DF, p-value: 0.5133

谢谢！

Answer 1

与大多数具有分类协变量的模型一样，第一级用作参考水平。在这种情况下，＆＃34;否定＆＃34; category被用作参考类别，因为默认情况下，R按字母顺序排序因子的级别。当您拥有分类数据时，您无法梳理出特定于人的平均值和参考类别的均值。它们被组合成截距项。那么CLASSneutral的系数不是neutral类的效果，它是neutral和negative的效果之间的差异。 CLASSpositive也是如此 - positive和negative的效果之间存在差异。因为模型默认使用单独的效果，每个人都有自己的拦截，我假设他们没有在摘要上打印它。

这不是plm独有的。使用标准lm也会发生同样的事情。

查看R中plm输出的所有估计值

1 个答案: