🎉 🎉 RAPIDMINER 9.10 IS OUT!!! 🎉🎉

Download the latest version helping analytics teams accelerate time-to-value for streaming and IIOT use cases.

CLICK HERE TO DOWNLOAD

"Neural Nets and Preprocessing Data"

chaosbringerchaosbringer Member Posts: 21  Maven
edited May 2019 in Help
Hi,
i have a question (of course)  ;D
Is data preprocessing for neural nets beyond normalization reasaonble?
In especially:
a) Can standardization be benefical?
b) Have strongly correlated attributes a negative effect on the net?
c) May PCA be adviseable?
d) My distribution is strongly inbalanced, having a few high values and a lot very very low values. Is that a problem?

Thank you very much.

Answers

  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,531   Unicorn
    Hi,
    to cut it short: 4 times Yes. But also: 4 times: Depends.

    In more detail: You really can't say what will help for your data. Just try it and you will know (if you take into account all needed considerations about estimating performance).
    You just can't say what will help without taking the data into account, you can't say which learner is better or which preprocessing. Everything depends on the data...

    Greetings,
      Sebastian
Sign In or Register to comment.