0
$\begingroup$

This web link is to a site that talks about forecasting building electricity, like a time series regression concept.

In the article they talk about the NN architecture as:

the architecture of this neural network can be written as 120:7:24

Is an MLP type NN? What I also dont understand is if they account for times series methods to forecast/predict. For example I thought for time series applications a sliding window concept needs to be used Vs a typical regression problem that does not have any element of time. Any tips greatly appreciated!

$\endgroup$
0
$\begingroup$

MLP typically refers to a type of neural network called a 'Multi-Layer Perceptron'. As you can read on the Wikipedia page, these neural networks consist of neurons organized in layers. Each neuron has a (typically fixed) function that transforms its weighted input to produce the output for that particular neuron known as its activation function.

A 'network architecture' refers to the way the neurons are laid out in layers in the network, sometimes including the activation function. In the given example, there are 120 neurons in the first input layer. These are connected to 7 neurons in the next layer, which in turn are connected to 24 neurons in the last output layer.

The 120 input neurons in the linked article appear to include both history, i.e. a sliding_window as well as some known future information (temperature, from weather forecasts it seems).

$\endgroup$
5
  • $\begingroup$ Hello, can you tell what type of Neural network used? $\endgroup$
    – HenryHub
    Feb 22 at 14:26
  • $\begingroup$ Would this type of code on the tensorflow Keras website for a regression example be applicable at all? Or is it not applicable because it doesnt account for time series type of a problem... tensorflow.org/tutorials/keras/regression $\endgroup$
    – HenryHub
    Feb 22 at 14:28
  • $\begingroup$ Hi, thanks for the question! To clarify: I am not the author of the linked article. The article does not mention the type of neural network. You would have to contact the authors or experiment yourself. $\endgroup$ Feb 22 at 14:31
  • $\begingroup$ Ill hit the green check box. Any chance you know much about if the code example in the tutorial for the tensorflow/keras would be applicable to predict electricity? Or should I be following time series prediction type ML tutorials? $\endgroup$
    – HenryHub
    Feb 22 at 14:39
  • $\begingroup$ Glad to have been of help, thanks for accepting the answer. The basic regression example you linked to in the Keras documentation does not account for the time series nature of the problem. I would encourage you to look into time series prediction methods for this. Some methods/tricks transform a timeseries problem into a regular regression by transforming the input. These methods may be compatible to the Keras tutorial. $\endgroup$ Feb 22 at 14:45

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.