I dispose of a data set composed of 6 features all of them are numeric and a binary target(taking 0 and 1). How should I proceed in order to predict the values of 2 features knowing the target and the remaining fixed 4 features.
Thanks in advance.
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
I am not sure what is the scale of your dataset, but I guess what you're looking for conceptually is Multi-task learning. You can use this concept to carry out multiple tasks, or in your case regression predictions using a single model.What you can try doing is use the four features, and the target variable as input features to predict the two variables.
If you have a large dataset, you can look into deep learning multi-taking approaches here. If you are familiar with python and sklearn, you can try some basic multi task regression algorithms available as a starting point. This algorithm might be helpful, its easy to use. If you are looking for a more deep learning approach, there are many architectures available online that you can try as starting point.
$\begingroup$
$\endgroup$
Target and features are relative to their usages in the Model.
You may treat 4 Features, 1 target as 5 features and 2 unknowns as target
Then predict one of the two targets using the 5 features
Then use that target as one more feature to predict the other target using 6 features
You may start with SGDRegressor
