In derived from the logistic regression model algorithm probabilistic model blog post, I talked about trying probability model from CHANG teacher curriculum model algorithm to derive a logical classification. Another fortunate enough to see a blog post 01 classification algorithms - Logistic Regression - Logit function , I learned to derive another classification model: the odds ratio is derived.
Odds ratio
First, understand what odds ratio: odds ratio (odds ratio) Odds , a way to measure the association between features in classification. Is the ratio of the probability of the event occurring and the probability of the event does not occur: \ (\ FRAC {P}. 1-P {} \) . Baidu Encyclopedia explained .
Logit model
We need to understand another concept Logit Model Logit function Baidu Encyclopedia explanation In addition this also explains some of the Logit is counted how? .
Related
We believe that the existence of a linear relationship between logit (odds) and the characteristic value X, namely: \ (logit (odds) = WX + B \)
further derive.
\ [Logit (odds) = ln (\ frac {p} {1-p}) = wx + b \]
\ [\ Frac {p} {
1-p} = exp (wx + b) \] provided
\ [z = wx + b \
] to obtain
\ [\ frac {p} { 1-p} = exp (z) \ ]
\[ p=\frac{exp(z)}{1+exp(z)}=\frac{1}{1+exp(-z)} \]
The main aspects of mathematics knowledge used: maximum likelihood functions , odds ratios , Logit model , basically knowledge of statistics, it appears that this will be the direction my next review and study.