Scaling the features for every model?



Scaling the features/variables is recommended for only cluster analysis or for any model building ?

if you were to run logistic model or any other classification regression with 5 independent significant features, where one feature has very high range compared to other. Would you need to scale the features ?


From my experience,

I ONLY scale features when I am using NeuralNet, KNN, or K-Means.
For Logisitc & Linear Regression, I don’t usually scale the features unless regularized.
Random Forest and Decision Trees don’t require the features to be scaled.