多项式朴素贝叶斯分类器如何处理类别特征?

时间:2018-12-09 15:52:08

标签: machine-learning classification probability naivebayes multinomial

我学习了如何使用MLE来获取朴素贝叶斯类别特征的似然发生概率。

watermelon quality prediction example:
  color texture quality
1 green clear   good
2 black clear   good
3 white blur    good
4 green blur    bad
5 black blur    bad
6 white clear   bad

我所知道的特征似然是MLE的P(纹理=清晰|质量=好)= 2/3。多项式分布如何拟合特征概率分布,然后预测P(texture = blur | quality = good)?

text type prediction example(bag of word, every feature is word count):
  pig dog cat type
1 3   4   1   1
2 1   2   3   1
3 4   5   1   0
4 1   1   1   0

据我所知,朴素贝叶斯分类器需要朴素贝叶斯分类器需要概率分布来拟合特征分布,例如高斯分布。多项式分布如何拟合特征分布P(pig | type = 1)?

如何使用拟合的多项式分布来预测P(pig = 0 | type = 1),P(pig = 1 | type = 1),P(pig = 3 | type = 1)?

0 个答案:

没有答案