Options
Generating time series attributes
Contributor II
Hi,
How does one generate custom time series attributes that uses past labels as inputs. I understand there are Pre Built processes like Moving Average but for more complex attributes, like, for example an oscillator:
((current example set value) - (low value of historical range (m ) ) / (high value of historical range (m ) - low of historical range (m) ))
How does one neatly do this such that variable M (the lookback range) is a parameter than can be optimized by one of the optimization processes. I believe one could potentially window data but that sounds awfully messy. These types of things are easy in MS Excel but I'm not sure how to approach this in Rapidminer.
Thanks,
Rob
How does one generate custom time series attributes that uses past labels as inputs. I understand there are Pre Built processes like Moving Average but for more complex attributes, like, for example an oscillator:
((current example set value) - (low value of historical range (m ) ) / (high value of historical range (m ) - low of historical range (m) ))
How does one neatly do this such that variable M (the lookback range) is a parameter than can be optimized by one of the optimization processes. I believe one could potentially window data but that sounds awfully messy. These types of things are easy in MS Excel but I'm not sure how to approach this in Rapidminer.
Thanks,
Rob
Tagged:
0
