Q1. What logistic regression will classify if z = w*x + b =0 , will it make that point as class 1 or class 0.
Q2. What if I use linear regression to predict the ypred and then normalize it between 0 and 1 and set the threshold value as 0.5, so if ypred is > 0.5 say it is class 1 else class 0. So without using logistic regression, I am classifying the points, is there any issue in this approach will it work or not?
Q1. it depends on how you set the threshold. if you identify a prediction as positive when the probability is >= 0.5, then having z = 0 means that the probability is equal to 0.5 and that means a positive.
Q2. I suggest you to watch the course 1 week 3 video “Motivation”. Perhaps you can start from 4:30.
Q1 is clear, Q2 I watched the video, I just want to ask will these work?
Find a dataset, try it, and compare how good linear regression vs logistic regression will work