Should we apply cuts just on the variable we plot and the weights, rather than the entire dataset to save some memory?