Is it possible to create a classification model which can predict continuous classes, like a number?
So far I've working with predictor which can predict one of two classes. I've searched about it and I just found references about continuous features.
I have a bunch of features (numbers) to predict a variable, namely 12 features, and observations labelled (with the variable to predict) that can be considered train data.
Best Answer
If the dependent variable is metric (continuous, and 1 means twice as much as 0.5), the model is called a regression model, not a classification model.
There are also models like logistic regression that are in between classification and regression: it is a classifier in that the classes are distinct groups, but model actually models the class membership probability which is metric and that's why it is logistic regression.
If the dependent variable still comes in groups, but they do have an order ("2" is larger than "1", but not necessarily twice as much), this is called ordinal regression.
If your problem has proper groups, but you encounter mixtures of the classes, that is yet another situation (regression again, but what kind depends on the actual situation/problem at hand).
I agree with @January that we need more information what your data is about to give you more detailed and sensible answers.
Similar Posts:
- Solved – How to find correlation if we have continuous and categorical variables present in the dataset as features and target is again binary
- Solved – Methods to reduce the number of class labels
- Solved – Classification of Huge number of classes
- Solved – Is there ever a reason to solve a regression problem as a classification problem
- Solved – Is there ever a reason to solve a regression problem as a classification problem