dont write code for this answer it logically

Question 1: Suppose that we have an unbalanced data for a data science problem, i.e., 1,100 instances, which is classified into the following distribution A=500,B=300, and C=300. Your task is to perform a detailed analysis of the given data. However, initially, you have to use a data split technique. What would be the best strategy to split the data? Perform a train/test split on the previously mentioned data. Justify your answer with arguments. Furthermore, specify what would be the total number of instances in each class if you do a class balance split