Tried SMOTE, which was supposed to be a relatively computationally cheap way to get the job done, but the computer kept crashing so I ended up completely uninstalling imblearn (the package used for resampling imbalanced datasets) and opted for regular balanced class weights within logreg. Next step is to check if a simple undersampling of majority class in target variable is computationally affordable.
Tried SMOTE, which was supposed to be a relatively computationally cheap way to get the job done, but the computer kept crashing so I ended up completely uninstalling imblearn (the package used for resampling imbalanced datasets) and opted for regular balanced class weights within logreg. Next step is to check if a simple undersampling of majority class in target variable is computationally affordable.