This article discusses how to generate enough test data to be useful for testing with the expected distribution of values and correlation between columns.