How should i select inputs and outputs for a power load forecasting problem by back propagation method in neural networks?

I have a set of data on real time power consumption with hourly granularity. I need to to use neural networks to predict future data. I need help on selecting inputs and outputs for the backpropagation model in neural networks. Thanks.

Answer 1

[Greg Heath]

Your choice of inputs and outputs are chosen by YOUR needs. Your choice of neural network and initial choice of network parameters are chosen, in most part, by the size and type of data.

What are your inputs and outputs?

How much data do you have?

[shapna]

Thanks for your help. I have data of hourly granularity for a year. It is real time data and it doesn't need pre-processing.I have weather data also linked to this for an entire year. now i would like to choose prediction in backpropagation method. I need to analyse this data based on seasons,weekdays,weekends and holidays. I would like to know how to select inputs and outputs for this scenario. Thanks.

Answer 2

[Greg Heath]

select outputs by want you want predicted

select inputs by using data that will help predict the outputs.

Greg

Answer 3

[shapna]

Thanks for your help sir. My inputs are date,hour, power consumption in mwh,and mean temperature.I need outputs that is the power prediction for next year. The targets i give in a matrix what should it contain sir? The values that i assume? Thanks.