If you have access to a MATLAB license with the Neural Network Toolbox, i suggest you start from there. It has a very easy Graphic User Interface system that let´s you choose the inputs, outputs, and different training scenarios.

http://se.mathworks.com/help/nnet/getting-started-with-neural-network-toolbox.html

See the tutorials for examples.

If you don´t have acess to this toolbox, there are some open acess (e.g. http://www.simbrain.net/).

I also suggest you to use MATLAB with neural Network toolbox and you can read the following papers:

https://www.researchgate.net/publication/273694673_Performance_Prediction_of_Solar_Collector_Adsorber_Tube_Temperature_using_a_Nonlinear_Autoregressive_Model_with_eXogenous_Input

https://www.researchgate.net/publication/276207167_Identification_and_control_of_inside_temperature_of_a_zero_energy_cool_storage_system_using_neural_networks

https://www.researchgate.net/publication/257018767_Dynamic_optimization_of_inside_temperature_of_Zero_Energy_Cool_Chamber_for_storing_fruits_and_vegetables_using_neural_networks_and_genetic_algorithms

https://www.researchgate.net/publication/272481612_An_Intelligent_Control_Technique_for_Dynamic_Optimization_of_Temperature_during_Fruit_Storage_Process

Article Performance Prediction of Solar Collector Adsorber Tube Temp...

Conference Paper Identification and control of inside temperature of a zero e...

Article Dynamic optimization of inside temperature of Zero Energy Co...

Article An Intelligent Control Technique for Dynamic Optimization of...

use K-fold , you can use 90% for train and 10% for test and so on also you use 99% for train and 1% for test , use time series algorithm or adaline

Thanks guys. Let me try. I have a license for MATLAB 2010