search

dmirman-zz / gazer

Functions for reading and pre-processing eye tracking data.
16 stars 14 forks source link

raw vs parsed data #11

Open hcuve opened 5 years ago

hcuve commented 5 years ago

Hi GazeR team,

First of all, really impressive work with the package and GCAs overall. I've been following your work and it is really a contribution to salute.

A questions I have for a while now and was never to able to find an exact response in your papers (especially the earlier work on GCA) is whether the models were fitted on raw data (using just raw gaze samples) or it was always done on parsed data (after running through an algorithm for fixation and saccades detection). If both cases are possible, has there been any systematic investigation to see whether the results are consistent both ways?

Thank you.

Best wishes, Helio

jgeller112 commented 5 years ago

Hi!

Thanks for the kind words! Right now we are just using the Sample and Fixation reports from SR. However, I am working on incorporating raw import of EDF files with blink/saccade/fixation event detection from the saccade R package to make everything more transparent and open. It will be available shortly :).

hcuve commented 5 years ago

Thank you for the response! so when you say samples, I take that it is fine to use raw samples without parsing, is it correct?

Cheers, H

jgeller112 commented 5 years ago

Use raw samples in what context? Every eye-tracking company has their own conventions when it comes to event parsing. The data ultimately needs to be parsed.

hcuve commented 5 years ago

Hi @jgeller112 , I was under the impression (at least based on some of the resources I found from your lab online) that some experiments just used raw gaze time series samples reported by the eye-tracker, without necessarily parsing it into fixation or saccades. And simply used that to get a sense of where subjects were looking at. I thought that the use of parsed data for GCAs seems to be a relatively new implementation in this package. When reading some of the previous work f of Dr. Mirman, as well as Dr. Barr, based on the descriptions, it seemed to me that they used raw gaze samples (meaning every frame reporting gaze position coordinates). Sorry I might just be confusing things a bit.