model2 is better, cover more data
binomial model p(1-p)
standard error
comuputing lower confidence level
Example
entropy越低越好， 代表信息的集中性
increase information gain, decrease entropy
what if inputs are numeric?
variance
gini
model2 is better, cover more data
binomial model p(1-p)
standard error
comuputing lower confidence level
Example
entropy越低越好， 代表信息的集中性
increase information gain, decrease entropy
what if inputs are numeric?
variance
gini