I want to generate synthetic test data by using the same features and structures of the training data. I want that it should follow the same types of values and range of values as of training dataset. I want it should not be having any missing values. it should follow the structure or constraints or relationships of traning dataset. is there any open source GUI available for the same where i specify or upload my tranining set and it will give me new test dataset? second a bit confused what size should i specify for this new dataset? same as 20% of training that what we usually do at the time of train:test split ratio?