如何在 glmnet 中绘制正确的标签？

Question

考虑这个例子

library(dplyr)
library(tibble)
library(glmnet)
library(quanteda)

dtrain <- data_frame(text = c("Chinese Beijing Chinese",
                              "Chinese Chinese Shanghai",
                              "this is china",
                              "china is here",
                              'hello china',
                              "Chinese Beijing Chinese",
                              "Chinese Chinese Shanghai",
                              "this is china",
                              "china is here",
                              'hello china',
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              'japan'),
                     class = c(1, 1, 1, 1, 1,1,1,1,1,1,1,0,0,0,0,0,0,0,0))

我使用 quanteda 从这个数据帧

得到一个 document term matrix

dtm <- quanteda::dfm(dtrain$text)
> dtm
Document-feature matrix of: 19 documents, 11 features (78.5% sparse).
19 x 11 sparse Matrix of class "dfm"
        features
docs     chinese beijing shanghai this is china here hello kyoto japan tokyo
  text1        2       1        0    0  0     0    0     0     0     0     0
  text2        2       0        1    0  0     0    0     0     0     0     0
  text3        0       0        0    1  1     1    0     0     0     0     0
  text4        0       0        0    0  1     1    1     0     0     0     0
  text5        0       0        0    0  0     1    0     1     0     0     0

我可以轻松地使用 glmnet 拟合 lasso 回归：

fit <- glmnet(dtm, y = as.factor(dtrain$class), alpha = 1, family = 'binomial')

然而，绘制 fit 不显示 dtm 矩阵的标签（我只看到三个曲线）。这里有什么问题？

Answer 1

据我了解，该图为您提供的是与重要单词相关的系数值。在你的例子中，单词 9-11 分别是 Kyoto、Japan 和 Tokyo（我可以从 dtm table 中看到）。这个普通的情节库没有我想你说的你想做的。相反，您可以使用 library(plotmo)，如下所示：

library(dplyr)
library(tibble)
library(glmnet)
library(quanteda)
library(plotmo)
dtrain <- data_frame(text = c("Chinese Beijing Chinese",
                              "Chinese Chinese Shanghai",
                              "this is china",
                              "china is here",
                              'hello china',
                              "Chinese Beijing Chinese",
                              "Chinese Chinese Shanghai",
                              "this is china",
                              "china is here",
                              'hello china',
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              "Kyoto Japan",
                              "Tokyo Japan Chinese",
                              'japan'),
                     class = c(1, 1, 1, 1, 1,1,1,1,1,1,1,0,0,0,0,0,0,0,0))


dtm <- quanteda::dfm(dtrain$text)
fit <- glmnet(dtm, y = as.factor(dtrain$class), alpha = 1, family = 'binomial')
plot_glmnet(fit, label=3)            # label the 3 biggest final coefs

干杯！

如何在 glmnet 中绘制正确的标签？

how to plot the correct labels in glmnet?

r

glmnet