machine learning 03 logistic regression

classification problems

logistic regression

try to solve the discret value
predict 2 values:0 Nagetive Class , 1 positive Class
Nagetive Class
the intuition : convey absence of something

how to develop a classification algorithm?

(1) linear regession :use it to classify

the question is that: sometimes it may give us a worse hypothesis when we add some examples
apply linear regression to a data set ,you may get lucky,but usually often it is not a good idea
another question is that : the hypothesis can output values much larger than 1 or less than 0,but the label is 1 or 0.it seems to be strange.

(2) Logistic Regression

0 <=h_θ(x) <= 1
a classification algorithm

Logistic Regression Model

(1) sigmoid function(logistic function)(2) interpretation of Hypothesis Output
which is meaning probability

(3) just like the linear question ,we should pick a value for the parameters theta