【2019.06.10】【学习总结】关于贝叶斯网络的应用，结合我的project实例

• 打个总结
  这个学习终于写完了最后的作业，可以说这个学期真的狠不容易了。从week2 开始到week13结束，基本每周都会有1~2due要交。还好大部分作业都拿了HD，D，所以final的压力会小一点。这个学期最难的两门课5047 AI 和5148大数据基本上搞得我心态爆炸。因为5047学习了naive Byes，但是并没有进行编程上的实现，所以在5147可视化的这门课大作业里，就异想天开想用Byes 来进行对LOL比赛的预测。第一次做，把代码贴出来，希望不对的地方还能指出。

首先对我的数据集做了清理，每个对位之间，谁能取胜都是独立事件，所以赋予每个对位信息“Blue"“Red”. 这里result 的意思是，蓝方获胜，no代表的是红方获胜。因为胜负也是独立事件。

根据贝叶斯公式：P（A | B,C,D,E) = P (B|A) * P(C|A) * P(D|A) * P(E|A) * P(A), 可以构建函数，代码如下：

prior.yes <- sum(train_set$result == 'yes') / length(train_set$result)
prior.no <- sum(train_set$result == 'no') / length(train_set$result)

bayespre <- function(condition){
  post.yes <-
    sum(train_set$TopWinningTeam == condition[1] & (train_set$result == 'yes')) / sum(train_set$result == 'yes') *
    sum(train_set$JungWinningTeam == condition[2] & (train_set$result == 'yes')) / sum(train_set$result == 'yes') *
    sum(train_set$MidWinningTeam == condition[3] & (train_set$result == 'yes')) / sum(train_set$result == 'yes') *
    sum(train_set$ADCWinningTeam == condition[4] & (train_set$result == 'yes')) / sum(train_set$result == 'yes') *
    sum(train_set$SupWinningTeam == condition[5] & (train_set$result == 'yes')) / sum(train_set$result == 'yes') *
    prior.yes;

  post.no <-
    sum(train_set$TopWinningTeam == condition[1] & (train_set$result == 'no')) / sum(train_set$result == 'no') *
    sum(train_set$JungWinningTeam == condition[2] & (train_set$result == 'no')) / sum(train_set$result == 'no') *
    sum(train_set$MidWinningTeam == condition[3] & (train_set$result == 'no')) / sum(train_set$result == 'no') *
    sum(train_set$ADCWinningTeam == condition[4] & (train_set$result == 'no')) / sum(train_set$result == 'no') *
    sum(train_set$SupWinningTeam == condition[5] & (train_set$result == 'no')) / sum(train_set$result == 'no') *
    prior.no;
  
  prob.yes = post.yes
  prob.no = post.no
  predction = ifelse(post.yes >= post.no,"Blue win","Red win")
  
  result <- paste("The prediction of winning team is:" ,toString(predction) , ".",
                     "The probility of True is:" , toString(prob.yes)  , ".",
                     "The probility of False is:" , toString(prob.no),".")

  return(result)
}