Dear Amiya,

If you are using the toolbox of ANFIS, you don't need to normalize data, however if the matrix dimension is important you can use min-max normalization.

Please see Q / A through ResearchGate posed by Sasha K. in August 16, 2015 using the following link:

-Question for ANFIS model, if I normalize the input and not the output ...

https://www.researchgate.net/.../Question_for_ANFIS_model_if_I_...

Also see Q / A through ResearchGate posed by Rohit Sharma in May 16, 2016 using the following link:

-How to normalize/de-normalize data Set for Anfis trainng ...

https://www.researchgate.net/.../How_to_normalize_de-normalize_...

Here are links and attached files to help you in solving problem.

-beginner - ANFIS network based on Sugeno model I - Code Review ...

codereview.stackexchange.com/.../anfis-network-based-on-sugeno...

-Advances in Data Analysis: Theory and Applications to Reliability ...

https://books.google.dz/books?isbn=0817647996

Best regards

Thank you sir @ M. Lafifi, instead of using min-max normalization, can i use zero mean and unity variance normalization. i.e (xi-average value)/standard deviation?

Thank you sir @ M. Parsapoor for your answer.

Thank you very much sir @ O.Taylan..Actually i am new this area, can you suggest me how to implement ANOVE? THANK YOU

Thank you sir @ O.Taylan for your valuable reply.

You can use Minitab or IBM SPSS. Both are good for such purpose.

thank you sir @ S.Chanda