Hello Kardi, Thanks for your tutorial and it helps me a lot. I am an engineering person and in my field I read a paper in which they chose 1000 as size of sample set to generate from single feature vector. Way to randomly generate is adding noise (with RMS of 5%). Do you know the basis for these selections? Are there any rules or basis for choosing generation size (min/max values) and methods/ways to generate random data. I look forward to hearing from you. Thank you in advance and regards.