在sql中的朴素贝叶斯计算

SELECT 
  class,
  COUNT(*) / SUM(COUNT(*)) OVER () AS prior_probability,
  AVG(CASE WHEN feature = 'yes' THEN 1 ELSE 0 END) AS feature_probability
FROM 
  dataset
GROUP BY 
  class;

SELECT 
  class,
  prior_probability * feature_probability AS posterior_probability
FROM 
  (
    SELECT 
      class,
      COUNT(*) / SUM(COUNT(*)) OVER () AS prior_probability,
      AVG(CASE WHEN feature = 'yes' THEN 1 ELSE 0 END) AS feature_probability
    FROM 
      dataset
    GROUP BY 
      class
  ) AS probabilities
ORDER BY 
  posterior_probability DESC;