uhm, quick question bro, talking about feature engineering, are there any instances where we will have to use only some specific features because you seemed to use most of them and how do we know if the features we used are too much or too little so that our model can generalize well.
Hi Samuel, yes most cases you may need to perform what we call "Principal Component Analysis or Dimensionality Reduction". However in my experience I mostly do this when my feature variables maybe above 30 or more.
You can also run model.feature_importances_ which will return the variables that contribute most to the prediction of your model, that way you know what variable are useful or useless. Hope this helps?
uhm, quick question bro, talking about feature engineering, are there any instances where we will have to use only some specific features because you seemed to use most of them and how do we know if the features we used are too much or too little so that our model can generalize well.
Hi Samuel, yes most cases you may need to perform what we call "Principal Component Analysis or Dimensionality Reduction". However in my experience I mostly do this when my feature variables maybe above 30 or more.
You can also run
model.feature_importances_which will return the variables that contribute most to the prediction of your model, that way you know what variable are useful or useless. Hope this helps?It does , Thanks 😊
It does , Thanks 😊