Hey I am still not clear with what overfitting means with respect to hyperplanes in SVM for binary classification when the margin is not maximum distance separating the two support vectors can you explain this in detail taking the example of credit card fraud detection with two classes genuine$($class$=0)$ and fraud$($class $1)$ and features distance from home, distance from last transaction, ratio to median purchase price, by$\_$chip,by$\_$pin,online,repeat retailer can you clearly show the scenario where the model is overfitting and what is meant by overfitting does it mean we are too srtict with the feature values that are not so important to judge the class label i$.$e$.$ the condition is too strict to classify a data point as a fraud transaction and hence label a fraud transaction as a genuine one can you give clear scenario with possibly a clear diagram when the hyperplane is margin maximizing and when it is not and overfitting and what i understood of sensitivity of less margin hyperplane is that if I change sligtly the feature values of the training dataset of the genuine points the hyperplane would change by a lot and require constant retraining again and again when new data added to data set whereas a margin maximimizing one the hyperplane would not change much as still this hyperplane is margin maximizing on adding slightly new data points to the dataset?