0
$\begingroup$

I have the power consumption values of a household for 2 year. To preserve privacy I am down sampling this data before sending it to the utility company. My goal is to get back the original data from the down sampled data. I am thinking I would need some kind of learning based predicting algorithm. My professor suggested Markov Chain model but I am not sure if that will work and don't know where to start from.

Any help on what directions I can take will be of great help. Thanks in advance.

Some additional info: I know the type of house, number of residents and pets, type and number of appliances.

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ Right now it's not exactly clear what you are asking. General advise may be too broad. What have you tried? Do you have a model for power consumption? Is it discriminant or generative? How do you want to learn the model's parameters? Point estimates like Maximum Likelihood/Posteriori or Bayesian approaches? If you could specify a bit more, we may be able to guide you in a better direction. $\endgroup$
    – Nicholas Mancuso
    Jan 23, 2015 at 21:50
  • 2
    $\begingroup$ In addition to the helpful comments from Nicholas Mancuso, another good direction might be to do some self-study to study Markov models and hidden Markov models (HMMs), and then try applying them to your problem and see if you're able to frame a more specific question. $\endgroup$
    – D.W. ♦
    Jan 24, 2015 at 1:34

0

Reset to default

Your Answer

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