I have data, bunch of measured temperatures, from which I should use ANFIS model to forecast temperatures. I am using MATLAB neuroFuzzyDesigner and it requires training data to have output column, what should that be?
Thank you
the MATLAB ANFIS toolbox could be used for developing MISO (multi-input single-output) model, so the output data should be single column (nx1)
From what I know about ANN , the first step towards using measured data is to prepare them in the way you want the analysis to be done. For e.g: the "bunch" of #s need to be put in a time-series . The reason is your fuzzy uses some sort of a gradient descent algorithm . For you to get a prediction , it is always a must that you give a sequence for proper estimation (say in the FIS output of the anfis( ) ) .
Trust me : data preparation is fundamental to all analysis .
Say for 24 hours then it would be: x(t-1) .
Based on data , you can set options to check for errors etc etc
- Badges
- Science topic
- Similar topics
- Engineering
- Instrumentation Engineering