How to prepare data for predictionESN?

[jollydragon]

My question is how to regulate the dimensions of "x_train", "y_train" and "x_test" for predictionnESN? They all should be numpy arries, correct? Should all the 3 arries be 2 dimensions? Shall the 2nd dimension only be "1"?

In my code, I reshaped both "x_train", "y_train" into (100, 1). But got the error message "ValueError: cannot reshape array of size 0 into shape (0,newaxis)" while fitting it.

I'm frustrated to resolve this issue after searching the internet as it looks like no other people has the issue. It's appreciated if you(Mr. Zimmerrol) can share some light.

[jollydragon]

By the way, is there any difference for predicting between directly after trained and after trained with validated? Can separating some data for validating besides for training actually improve the predication performance?
Are there anything else about EasyESN I should learn to improve its predication performance? Where can I get it?

[darreal44]

The only advice I can give you is to take some machine learning courses online. Everybody can use a machine learning library but it's the way you prepare, validate, parameter or choose your data or your metrics which make all the differences between something useful and something which is not. Even with the same used algorithm you can get totally wrong results than another person who do things properly with the same tools. That's not easy to explain but hopefully there are plenty of way to learn it online.