machine learning problem.

Suppose that the local power company wants to predict electricity demand for the next 5 days. They have the data about daily demand for the last 5 years. Typically, the demand will be a number between 80 and 400.

a) Describe how you could use an Multi Layer Perceptron to make the prediction. What parameters would you have to choose, and what do you think would be sensible values for them?

b) If the weather forecast for the next day, being the estimated temperatures for daytime and nighttime, was available, how would you add that into your system?

c) Do you think that this system would work well for predicting power consumption? Are there demands that it would not be able to predict?