2017-07-11 - Issue with Project Discovery Evaluation Set

around 350ish, sometimes higher, sometimes lower, depends on the difficulty. I don’t get the easy ones, nor do I get “no transit” controls anymore.

Wait what? You can’t even spam click through it with “no transit”? How much EP do you get for one delivered result at 95% accuracy?

At 95% accuracy you get 87 EP.

So for 350 EP-ish a minute you have 15s for each sample. Only clicking all necessary buttons and “no transit” takes 9-10s already. Okay if you manage to do the rest in 5-6s, sounds stressful. Still 350 EP/min will take ages too

I think cases like the could be solved with an analysis accuracy check: let’s say if you marked 80% of the transits correctly in the sample then it counts as successful. Or partial success with reduced, but still positive rewards.

1 Like

It will say failed, but if you have enough transits marked your accuracy should go up.

I may be wrong, but this whole thing seems like a ploy to get us to use our cpu’s to process the algorithms to analyze the graphs and the “game” is just us guessing at whether we agree with the computer. It seems like there are only some of the plates that’s really take the users inputs while others tell you whether you are right or wrong. Well, for it to tell you you’re wrong, the computer would have had to already computed the transits. If that’s the case, then WE aren’t contributing anything, only lending our computer processing power to CCP and the University and then being told we are idiots for not seeing something that HAD to be computed via mathematical computations based on statistical data points.

Just saying, if I’m not getting the bitcoins from the mining, at least make the prizes commencerate with the CPU farming I put in and not based on some random guesses where the elusive transits are.

:psyccp:

Or maybe it is a social science study, finding out how much we are gonna take. But I like your conspiracy theory better. +1

2 Likes

Got this one twice in the 10 daily double XP samples on one of my characters. At least the second time I knew that I should not mark the zoomed part as a transition. Don’t know why though, because it sure looks like one.

Here’s my version of the same thing:

1 Like

The reason for that would be that, in projects like this, they need to know if people are just spamming away buttons or not. The ones where they tells us right/wrong are based on already computed data, they are control points.

The ones showing consensus statistics are actually the ones where we are contributing.

But hey, I love a good conspiracy theory! [tinfoil cap ON]

1 Like

The “analysis failed” and “Analysis successful” reports are just data for quality assessment ?

Don’t quote me on this, as I have no certainty. But it’s what makes sense to me.

You can quote me: yes, they are.

It probably is, but there aren’t any double planet controls.

actually you can but you need some knowledge about statistics and some experience in data analysis. it is not easy … but hell i like a good challenge!
Make sure you switch the detrend on and off ( 24 h works for me the best ) and look at the curves. the dimming you selected (red) is quite misleading. good job!

there is another dimming there. and yes you can add multiple planets . before submitting, you can start again and add another planet

Attention all capsuleers: scientific research is not easy ! :sunglasses:
some will get a hint on the tasks the researchers face in analyzing these samples. I have to add that also they have more tools!
I suggest

  1. Progressive sample difficulty
  2. CCP provides more documentation on PD and methodological tips for analysis!
  3. Add star information :
    Location,
    Age
    Chemical composition
    Diameter
    Kinematics
    Magnetic field
    Mass
    Rotation
    Radiation (1 Luminosity and Magnitude )
    Temperature
    Classification
    Variable star type (Pulsating,Eruptive, Cataclysmic or explosive variable eclipsing binaries, as well as rotating stars that produce extreme star spots) eg: Algol an eclipsing binary star regularly varies in magnitude from 2.3 to 3.5 over a period of 2.87 days.

If the Analysis Successful and Analysis Failed are for Quality Control Assessments then then would it not be better to at least provide a ‘Partially Successful’ standard where the Analysis succeeds on at least one transit but does not find others (i.e. you match at least one of the transits but do not match or see others that are in the sample). This at least would give those attempting to help in this some ‘heartening’ news and not deter them from just turning away from the PD in it’s entirety.

Good this and other tweeks, such as the zoom function, its not intuitive how to zoom.