Timeline for Smoothing 2D data
Current License: CC BY-SA 3.0
25 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 10, 2013 at 20:43 | answer | added | DarenW | timeline score: 1 | |
| S Feb 4, 2013 at 22:06 | history | bounty ended | CommunityBot | ||
| S Feb 4, 2013 at 22:06 | history | notice removed | CommunityBot | ||
| Feb 1, 2013 at 4:04 | comment | added | baptiste | btw, since I won't be connected when the bounty period expires, I'll leave it to default to the most upvoted answer. thanks. | |
| Jan 31, 2013 at 11:52 | answer | added | scellus | timeline score: 4 | |
| Jan 31, 2013 at 11:05 | answer | added | cbeleites | timeline score: 2 | |
| Jan 31, 2013 at 9:11 | answer | added | zorbar | timeline score: 0 | |
| Jan 30, 2013 at 21:14 | comment | added | baptiste | there's no audio signal, it's a series of optical spectra. I've added a dummy picture for illustration. | |
| Jan 30, 2013 at 21:12 | history | edited | baptiste | CC BY-SA 3.0 |
added 300 characters in body
|
| Jan 30, 2013 at 11:22 | comment | added | zorbar | @baptiste: I get it; so is an audio spectrogram of voice: it's not an image; but looking at the image certainly helps, and sometimes you can actually use image-processing techniques to eliminate noise. Did you look at it as an image? can you display your data as a spectrogram, meaning: the horizontal axis is the time-axis; the vertical axis is the frequency axis; and in a given time you show the frequencies by coloring them by amplitude? [see wikipedia explanation if you're not familiar with it] | |
| Jan 30, 2013 at 10:55 | comment | added | baptiste | you can certainly display the data as an image, or think of it as an image if that helps, but the physical acquisition process is a sequential recording of optical spectra. | |
| Jan 30, 2013 at 9:51 | comment | added | zorbar | @baptiste, please correct me if I'm wrong, but it sounds like you're describing images, right? And since a picture is worth a 1000 words, perhaps it's best if you upload a sample image or some images. Then it can be easier to look at and explain the problem; I also suspect that perhaps a computer-vision / image-processing solution may be appropriate [which is my expertise] | |
| Jan 28, 2013 at 13:17 | answer | added | user1966337 | timeline score: 1 | |
| S Jan 27, 2013 at 21:01 | history | bounty started | baptiste | ||
| S Jan 27, 2013 at 21:01 | history | notice added | baptiste | Canonical answer required | |
| Jan 26, 2013 at 23:55 | comment | added | baptiste | in other words, if I am to treat these data as time series, is 'time' going to be the actual time (x dimension), or could it be the optical frequency (y dimension)? | |
| Jan 26, 2013 at 4:02 | comment | added | baptiste | is there something specific to the time variable that makes time series a particular type of statistical analysis? | |
| Jan 25, 2013 at 22:39 | comment | added | user12719 | I'd look into 'robust' methods. These methods try to de-weight outliers. E.g. there is a robust spline algorithm in R. | |
| Jan 25, 2013 at 22:03 | history | tweeted | twitter.com/#!/StackStats/status/294928464098766848 | ||
| Jan 25, 2013 at 21:40 | comment | added | Peter Ellis | My hunch is to treat them as 500+ correlated time series and use time series techniques like moving average or exponential smoothing; and only use 2d smoothing afterwards and only if necessary for a stylised graphical representation. I don't have enough backing this up however to turn it into a proper "answer". | |
| Jan 25, 2013 at 21:32 | history | edited | Peter Ellis |
added the time-series tag as I think this is key to the problem
|
|
| Jan 25, 2013 at 21:23 | comment | added | baptiste | @PeterEllis a large number (say 500, but for the sake of generality it could be even larger) | |
| Jan 25, 2013 at 21:21 | comment | added | Peter Ellis | How many frequencies were measurements taken at? If it's not a large number, might it be practical to see this as a set of individual (but related) time series, one for each specific frequency? | |
| Jan 25, 2013 at 21:00 | review | First posts | |||
| Jan 25, 2013 at 21:03 | |||||
| Jan 25, 2013 at 20:40 | history | asked | baptiste | CC BY-SA 3.0 |