使用预测与朴素贝叶斯一起。能够计算条件概率但产生完全错误的预测概率

Question

我正在尝试预测是否播放给定的新数据：

m <- NaiveBayes(as.factor(play)~ ., data=train)
weather     Temp    humidity    wind    play
fine        hot     high        none    no
fine        hot     high        few     no
cloud       hot     high        none    yes
rain        warm    high        none    yes
rain        cold    medium      none    yes
rain        cold    medium      few     no
cloud       cold    medium      few     yes
fine        warm    high        none    no
fine        cold    medium      none    yes

我用它来创建新的测试实例

`> wind <- factor("few",levels(train$wind))
    > Temp <- factor("cold",levels(train$Temp))
    > humidity <- factor("high",levels(train$humidity))
    > weather<- factor("cloud",levels(train$weather))
    test <- data.frame(weather,Temp,humidity,wind)
    predict(m,newdata=test)` 

这是输出

$class
[1] yes
Levels: no yes

$posterior
              no       yes
[1,] 0.004836495 0.9951635

我也试过

m <- NaiveBayes(as.factor(train)~ ., data=train.play)

但返回相同的概率