CODEWITHSUNDEEP
pythonregressionclassification
Muhammad Asad
Muhammad Asad

Reputation: 55

Python: Calculate score with binary target data

I have some data giving whether or not some influencer is a bot. Of course the target data is in binary values. I need to use this whole data to calculate a score for the influencer rather than a direct prediction. So here I am stuck with three options with no definite answer that I could find for which one to use. The data is obtained from Instagram API and I am to experiment with some features or pick some from work of others.

  1. Should I train a classifier and remove sigmoid function from the last layer to get continuous values and use as a score; is this possible to do in python?
  2. Is there a way to train a regressor with the binary target data and would that be suitable?
  3. Or am I just left with an option to create a formula like that of engagement for the score with what information I have?

Please guide as I am a newbie and now completely stuck.

Upvotes: 0

Views: 197

Answers (1)

CutePoison
CutePoison

Reputation: 5365

This exactly what a Logistic regression does - it calculates the probability of an input being class 1.

Say you chose your labels to be [bot,no-bot] = [1,0] then it gives you the probability of an input being a bot.

Upvotes: 2

Related Questions