Question

vocab
 wordIDx V1
    1  archive
    2  name
    3  atheism
    4  resources
    5  alt

wordIDx newsgroup_ID    docIdx  word/doc    totalwords/doc totalwords/newsgroup wordID/newsgroup    P(W_j)
1   1   196 3   1240    47821   2   0.028130269
1   1   47  2   1220    47821   2   0.028130269
2   12  4437    1   702 47490   8   0.8
3   12  4434    1   673 47490   8   0.035051912
5   12  4398    1   53  47490   8   0.4
3   12  4564    11  1539    47490   8   0.035051912

对于vocab中的每个wordIDx，我需要计算以下公式：例如wordIDx = 1; 我的值应该是

max(log(0.02813027)+sum(log(2/47821),log(2/47821)))
= -23.73506

我现在有以下代码：

 classifier_3$ans<- max(log(classifier_3$`P(W_j)`)+ (sum(log(classifier_3$`wordID/newsgroup`/classifier_3$`totalwords/newsgroup`))))

我该如何循环使用，以考虑到vocab数据帧中的所有wordIDx并计算上面突出显示的示例。

Answer 1

类似的事情，但是您确实需要清理列名。

vocab <- read.table(text = "wordIDx V1
1  archive
2  name
3  atheism
4  resources
5  alt", header = TRUE, stringsAsFactors = FALSE)

classifier_3 <- read.table(text = "wordIDx newsgroup_ID    docIdx  word/doc            totalwords/doc totalwords/newsgroup wordID/newsgroup    P(W_j)
1   1   196 3   1240    47821   2   0.028130269
1   1   47  2   1220    47821   2   0.028130269
2   12  4437    1   702 47490   8   0.8
3   12  4434    1   673 47490   8   0.035051912
5   12  4398    1   53  47490   8   0.4
3   12  4564    11  1539    47490   8   0.035051912", header = TRUE, stringsAsFactors = FALSE)

classifier_3 <- classifier_3[!duplicated(classifier_3$wordIDx), ]
classifier_3 <- merge(vocab, classifier_3, by = c("wordIDx"))
classifier_3$ans<- pmax(log(classifier_3$`P.W_j.`)+ 
                     (log(classifier_3$`wordID.newsgroup`/classifier_3$`totalwords.newsgroup`) +
                            # isn't that times 2?
                            log(classifier_3$`wordID.newsgroup`/classifier_3$`totalwords.newsgroup`)),
                        log(classifier_3$`wordID.newsgroup`/classifier_3$`totalwords.newsgroup`))

从一个数据框中获取值以循环到另一个数据框中进行操作

1 个答案: