Project discovery - dodgy results


(Tara Stargazer) #1

This is ridiculous…


How could i have missed that?


(Tara Stargazer) #2


These 2 transits were a beautiful match when folded.

I gave up on PD a while ago because of obvious WTF moments like these.
Sure i had a few ohhhh… duh how did i miss that one moments too, but these are clearly false positives / negatives.
I’d have thought that these would be more carefully dealt with by now.


(Sindara T'Soni) #3

The thing is, it’s based on consensus, so if everyone clicks just the one point on the graph (as in your second example) then that becomes the ‘right’ answer, even if it isn’t.

:upside_down_face:


(Tara Stargazer) #4

That’s how i understand it.
Thing is, with such an obvious transit how could the consensus be that it doesn’t exist?
There should be a reporting feature for errors like this.
I’ts pretty bad when one so bleeding obvious gets missed.


(Waylon Dowd) #5

It’s the ones like this that drive me nuts…


(Tara Stargazer) #6

Not super impressed with those, but unless i’m mistaken because you have the majority on target i think you still get + accuracy.
At least i thought i saw that i did when i had one of those.
Something that may help, and apologies in advance if i’m telling you something you already know, re-positioning the mark in folded view so as is is centered after you’ve matched the transits up as best you can will often eliminate the issue you had in the above sample… but not always, especially for large numbers of transits over a short period.

Came back to this thread because of this sample…


The deviation on the right is a glitch in the sample, but the red is clearly a transit!


(Waylon Dowd) #7

The one I posted is a small example. I had one earlier today that had 15 out of 16 lined up, but one didn’t. And you’re right, you still get partial credit on accuracy. I just don’t know that I would call 15 correct out of 16 a failure.


(Nevyn Auscent) #8

No it isn’t.
Those samples are done by the actual scientists responsible for the project. They learnt about the consensus problems early on in the cell PD and removed it, now it’s just the official slides.

Now in some cases the scientists may have it wrong, or there may be some reason why it’s not actually a transit, because they have access to a lot more data to determine transits than we do.
It does suck when examples like the above come up, but the more people who hit them like that the more chance they’ll look at those slides. They cleaned it up a lot from the starting days.

Additionally make sure to use the non detrended view as well.


(Sindara T'Soni) #9

Correct. Not exactly sure how it works, or if it works every time, but you can definitely get accuracy increases for being mostly right.

There are still ‘no consensus’ results which doesn’t fit the idea that the only transits presented to us have already been officially identified by the scientists. Some of the ‘no consenus’ results have very clear transit patterns that fold perfectly, but it seems to be the case that until enough people have identified those transits, PD doesn’t credit you with a successful analysis.


(Tara Stargazer) #10

Here we are again…
This one folds nicely…


#200226162


(Nevyn Auscent) #11

Different magnitude.
Not saying I wouldn’t have also marked it, depending what the un-detrended view looked like also. But it’s not part of that particular pattern at the very least.


(system) #12

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.