search

sta-363-s20 / community

Discussion, Q&A, everything you want to say, formatted nicely
1 stars 0 forks source link

Maximum Likelihood in logistic regression #17

Closed Cindyyyhey closed 4 years ago

Cindyyyhey commented 4 years ago

Hi Professor McGowan,

"In logistic regression, we use maximum likelihood to estimate the parameters. This likelihood give the probability of the observed ones and zeros in the data"

I'm not quite sure what does the likelihood means here? I think the probability of observing 1 and 0 as response variables in the data is fixed.

LucyMcGowan commented 4 years ago

It's true the observed data is fixed - what maximum likelihood is doing is trying to fit a particular distribution to the data and this distribution has parameters (in this case some of the parameters are the beta coefficients in the model) - the idea is that the beta coefficients are chosen such that the probability of observing the data that was observed is maximized.

Cindyyyhey commented 4 years ago

thanks!! got it