"generating attribute with time series data ERROR"

tonio09

I'm trying to generate an a trend attribute.

unixtime, price, amount
1,3,5
2,5,7
....

I apply windowing, which creates these new attributes
price-0, price-1, price-2 ...

Next I try to generate a trend attribute so I use this expression
if(label>price-9,1,if(label==price-9,0,-1))

Pronlem is Rapidminer interprets price-9 as price minus 9, instead of the variable name. I can't change the variable names, because they are generated by RapidMiner in the windowing. How can I fix this?

MariusHelf

Heya Tonio,

you can change the attribute names with the Rename by Replace operator. Just enter something like '-' into the replace_what parameter, and '_' into replace_by.

Best,
Marius