I am working on situation where I have more than one independent ( Feature) variable, let say four and one dependent (targeted) variable, my confusion is when XGBoost algorithm applied on such data set, during prediction it will consider all feature variable as individual like linear regression do or as a group to develop decision trees (Prediction).
If any reference paper or book available related to supervised Machine Learning Algorithm , Kindly share
The answer is no. XGBoost assumes the independence of the predictors but uses their possible interactions as a basis to create an additive and surrogate structure made up of weak learners. Therefore, the effect of each predictor will be influenced by the effect of the other predictors and their number, and that is the extra point of boosting compared to conventional regression methods. To understand more deeply how XGBoost works, nothing better than a detailed explanation from its creators: https://doi.org/10.1145/2939672.2939785
Best regards!
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing