1
$\begingroup$

I have coincidence factor for different sizes of groups and the associated attributes (e.g. building usage type and floor area) for each consumer in the group. I want to predict coincidence factor for new groups using the attribute for each consumer in the new group and the number of consumers in the group. What machine learning method can solve the problem? :)

Defintions: Coincidence factor is the peak of a system divided by the sum of peak loads of its individual components

$\endgroup$
3
  • $\begingroup$ Would you show the regressors? are regressors somehow correlated to thi incidence factor? $\endgroup$
    – 3nomis
    Commented Feb 6, 2019 at 11:00
  • $\begingroup$ Yes. The coincidence factor fall with the number of consumers. The coincidence factor also depends on the building types e.g. private home use domestic hot water at the morning creating a spike which not existing for office buildings. $\endgroup$
    – Martin V.
    Commented Feb 6, 2019 at 11:16
  • $\begingroup$ I should mention that the output should be nummeric (e.g. between 0-1) - not a cateogry/class $\endgroup$
    – Martin V.
    Commented Feb 6, 2019 at 11:45

1 Answer 1

0
$\begingroup$

From what you are saying I guess you have both categorical and continuous variables, One suggestion would be to check the distributions of your features to see if, more or less, the criteria for a linear regression are met. If you have too many variables, which I still haven't picked up, you could even apply a PCA and then a regression.

$\endgroup$
2
  • $\begingroup$ What about supervised or unsupervised machine learning (clustering, SMV, ...) can they be used to solve the problem? $\endgroup$
    – Martin V.
    Commented Feb 6, 2019 at 12:46
  • $\begingroup$ What I just said 'Regression' is supervised learning. You don't need unsupervised as you already have a target variable defined. You could use a dimencionality reduction method(PCA) to ease the regression somehow $\endgroup$
    – 3nomis
    Commented Feb 6, 2019 at 13:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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