2
$\begingroup$

I am using the famous Titanic competition from Kaggle as an example. The binary classification problem is looking for a good classifier or a combination for predicting "survived" or not.

What if I have another data column as "survival rate" (numerical) in this problem, which of course could be easily converted to the categorical response of "survived". e.g., categorical "0" is corresponding to the survival rate "0 - 50%), and "1" is corresponding to the rate "50% - 100%". My question is really how the "survival rate" set could be helpful for the goal of classification.

My real world case happens to be a multi-class classification problem, which also comes with a corresponding numerical "rate" response. Thanks for any suggestions in advance!

Improve this question
$\endgroup$
0

2 Answers 2

Reset to default
1
$\begingroup$

If you have "survival rate" column, you have the extra flexibility of using Regression models as well as Classification models. Once predicted-"survival rate" is calculated, convert them to classes. This does not mean that Regression model will give a better model, just that you will have more model to try out.

Improve this answer
$\endgroup$
0
$\begingroup$

My 2nd thoughts is as below. Split the train into train1 and train2. Develop a regression model using train1 plus survival rate. The model is then used to create a new predictor "survival rate" for train2 and test set. Finally, a classification model is built for predicting "survived". As you can tell, this strategy is still not using all information provided. So any other idea?

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.