I would like to replicate a sample from a population. For this, the distribution (PDF) of $n$ characteristics are calculated. Then I just draw at random from each PDF, and put it as part of the new sample.

How this process is called (if it has a name)?

  • $\begingroup$ Depending on how the PDF is calculated, how many samples are obtained, and what is meant by "take note," this process is an integral part of many procedures, including parametric bootstrapping, multiple imputation, and Monte-Carlo simulation. $\endgroup$ – whuber Feb 4 at 16:53

If I understand correctly, you want to generate new samples based on a PDF determined from your data. Such a process is known as oversampling and there exists a paper which does exactly this. Check it out: https://ieeexplore.ieee.org/abstract/document/6252384

| cite | improve this answer | |
  • $\begingroup$ I'm reading the paper. If I understand, it is one PDF calculated for all the feature space? $\endgroup$ – Tomás Concepción Miranda Jan 31 at 15:16
  • $\begingroup$ Because it is not just one PDF, it could be $n$ features; so, I think, there will be also $n$ PDFs for each feature. $\endgroup$ – Tomás Concepción Miranda Feb 3 at 8:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.