Interpreting SAXContainer getOffsets() ?

[Tim Schultz]

Greetings all.

Quick question - how exactly am I to interpret the array returned by the getOffsets() member of the SAXContainer class in determining the start/end boundary points?

EX:

#1: [30, 90, 170] 

or 

#2: [10, 15, 78, 99]

Does this mean in #1 that the motif starts at 30, ends at 90, starts at 90, ends at 170?  How exactly am I to infer boundaries from this with single offsets?

#2 has an even number of offsets - does this mean start=10 end=15, start=15 end=78, start=78 end=99, etc.?

Maybe I'm over thinking this and looking at it the wrong way.

Appreciate the help.

Regards,

Tim

[Pavel Senin]

Hi Tim: 

I would guess you are asking about Sequitur-related code. We took this code offline for a while, - we work on proper implementation. If you could send me PM, we can discuss this issue further.

My current code doesn't have these methods anymore, but I think these could be starts of motifs/discords, while ends are [start + sliding_window_length], or these could be positions on the SAXString, which was built using the sliding window - then you'll need to figure out the position on timeseries on your own at first and add the sliding window length to get the end. 

I cannot tell you exactly because I worked on that part and had both variants implemented.

--
You received this message because you are subscribed to the Google Groups "jmotif-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jmotif-discus...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
Mahalo, Pavel.

[Tim Schultz]

Hi Pavel,

Appreciate the reply.  Yes, I am referring to the Sequitur code -- I did notice today that I no longer saw it on the wiki.  

By "proper implementation" do you mean refactoring the Java API, or fixing bugs and issues?  I seemed to have some success with Sequitur looking at data from various sources, so I'd be curious to know.

I'm actually a Ph.D. student as well, at Drexel University (healthcare informatics).

Should I reach out to you at your hawaii.edu email address?

Regards,

Tim

[Pavel Senin]

Yes, senin at hawaii.edu is valid. In fact, I am re-implementing logic around Sequitur core to be able to find those positions precisely and working on GUI right now.