Time series Labeling

Contributor II

Time series Labeling


Does anyone know how I can generate a new binary attribute based on whether the next value in a time series stock price is UP or DOWN?

For example, If I have

time1: 50
time2: 60
time3: 40

Then, the generated attibute would be

time1: UP
time2: DOWN
time3: ?

Thank you in advance

Regular Contributor

Re: Time series Labeling


The operator LabelTrend2Classification does exactly this; but if your attribute is not the label you might need to designate it as such, using the ChangeAttributeRole operator, apply the LabelTrend2Classification operator, and then change back to the original label.

Actually there is also a ChangeAttributeType operator which appears to do the same as the ChangeAttributeRole, so two for the price of one!
Contributor II

Re: Time series Labeling

Hello Haddock,

Thank you for your response. I was wondering if there is a way to set a threshold, so the trend defition is set only if the change in values is above this threshold.

Do you know if this is possible?

Best Regards,

Regular Contributor

Re: Time series Labeling

Should tell you in the documentation.
Contributor II

Re: Time series Labeling

I could not find anything related to threshold definition in this operator.

Does anyone know how to do this?

Elite II

Re: Time series Labeling

Hi Andre,
did you already applied a windowing function onto your time series? Then you could use feature construction and a UserBasedDiscretization to fullfill your goal. Lets say you have something like that as an example

time-3  time-2  time-1 time-0
30        40      45      52

You could then construct a new attribute difference using time-0  - time-1.
time-3  time-2  time-1 time-0 difference
30        40      45      52        7

Using a UserBasedDiscretization you could wrap everything between -5 and 5 to stable and the rest onto down or up.

Old World Computing - Establishing the Future

Professional consulting for your Data Science problems