2

I am preparing 2/3 neural networks models (I have chosen "neuralnet", "nnet", "AMORE" randomly among the R packages) for a certain problem. At the moment I do not have the Data Set, but I would like to train my neural networks on a similar one.

I am looking for a Data Set with some features:

  1. a set of n-1 independent variables (where n is not extremely large);
  2. a dependent variable assuming values in a very wide range (for example y varies from 0 to hundreds of thousands).
  3. Only a very few rows have a very high y.

Do you know some Data Set similar to the one I will have to deal with?

migrated from datascience.stackexchange.com Oct 19 '15 at 7:56

This question came from our site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field.

  • That is a very specific data set. The answer to this question would not help anyone except you. This is not the purpose of the site. Is there any way you could re-phrase your question so it is not about your specific data set. For instance, what has prevented you searching for a similar data set, have you looked at any sources of machine learning tutorial data (such as archive.ics.uci.edu/ml/datasets.html )? Have you tried simply constructing a test dataset over a mathematical model? – Neil Slater Oct 18 '15 at 18:46
  • 1
    First of all thanks for the site (I hope this site is not only for already experts in Data Science, otherwise... why this site?!?!). Second, I tried to write it more generally (I had given numbers to help the answers). Third.... I belong to other StackExchange sites: I had given and received a lot of aids, without overthinking. Maybe some data scientist has worked on a data set similar to the one I'm looking for, so... maybe he will be happy to share his knowledge with me and the community. Anyways, thanks for the advices. – Andrea Ianni ௫ Oct 18 '15 at 19:22
  • I'm voting to close this question as off-topic because it depends on theOpenData SE – Dawny33 Oct 19 '15 at 5:07
  • Stupid question probably, but why can't you just simulate a data set to your requirements? Given there's no point at all in training a NN on a dataset that isn't the TRUE training data for your real dataset, you can't actually do anything useful on your true data with any other dataset (unless you tell us what it is, like pictures of tanks). So my stupid question is basically saying "do you really just want to find out how the R packages work?" – Spacedman Oct 19 '15 at 7:33
1

The Data Set with the required features I have chosen at the end is:

"ForestFires"

You can download it here.

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.