Logistic regression


In logistic regression, we trying to get a final outcome which is either,

a) TRUE / FALSE

b) Between 1 to 0

Perhaps the following equation would paint the picture a little better.


y = 1 / e^- (a + b1x1 + b2x2 + x3b3...)

As you can see from the equation above, we will be getting a result  0 to 1 or true / false, depending on how you calculate it.

We will be using our data earlier and call a very important function in R, glm with family parameter set to binomial as shown below :-





And the result is as follows. Here can can see that degree is a strong influence of whether a person gets high salary or not.




Residue deviance - is lack of fit of your model taken as a whole.

Fisher scoring iteration is to say how a model is estimated.


As for the interpretation of results, here is one great link.




Comments

Popular posts from this blog

OpenCover code coverage for .Net Core

ionic2 cordova build android - Unable resolve gradle 2.2.3

Using Custom DLL with IronPython / Scripts