To my eyes 50% of the calibration controls (where you get increase or decrease of your hit rate %) are really really hard to understand. They all have 1-2 d orbit times and the outlined dot pattern distributes evenly around the average of the curve with no visible dip or pattern. It is sad that you cannot redo analysis and folding on these examples afterwards, because I would be confindent to learn something from the fold of these samples and get a better unterstanding what to look for (at least from some of them, other appear to be pure noise).
aside from the fact that molten mercury type planets (<2 d orbit) are not overly exciting, these controls are really frustrating because one doesnât see a pattern even if it is highlighted in red âŚ
During phase 1 one project discovery I could analyse lots of samples without getting tired.
When I failed a control sample, I could see where why I failed after getting the supposed right answer.
These new phase of project discovery is very tiresome, consumes a lot of energy from me.
I think a lot of the stress is due to not being able to learn from failed control samples. Even after folding the supposed right answer there is no visible pattern, no clue to look for. Looks like the lvl 4 or lvl 5 control samples were part of a bigger data set and that bigger data set had clues that leads to the transients, but those clues are missing in the smaller samples we receive.
I donât mind the samples being hard, as long as I can learn when I fail. That is not the case right now.
After analyzing several hundred more samples, I thought of another feature that I feel would be very useful: Iâd like to be able to âmow the grassâ, removing any data points that are exceptionally above average. For one thing, it would eliminate problems with the graph scaling too much on very active stars. The main benefit, though, is that it would eliminate more than a few false positives, where an otherwise unremarkable wiggle in the graph appears much more significant than it really is because itâs surrounded by a pair of solar flares. Itâs also possible that such a data culling feature would make it harder to spot something, but knowing when to use it will be up to the analyst. More tools canât hurt, right?
EDIT: I just had a sample (Not a gold standard grading sample) with a period of 0.475 days. Since the UI only allows periods as low as 0.5, I had to enter it as two different orbits. Iâm sure itâll still get examined by Geneva, so op success, but itâs a bit annoying.
After reaching rank 10, i have a request regarding the âfailed analysis samplesâ. Would i be correct in assuming these are prevously researched samples that are used to establish our accuracy score? Some of these when marked do not look like a transition to me. So to increase my abilities to recognize these far more subtle transitions, i would like to be able to mark and fold over the points that are highlighed that i missed. Then i can manipulate the data until i also see the transition. Doing this on the data sets i fail on will increase my abilities to recognize the patterns. I would also like a calliper like used for EKG interpertation, to measure gaps/duration of transitions and then compare to the rest of the data set for matches.
What actually irritates me the most (aside from the random results) are the BPC.
I have quite a few ships (Victorieux Luxury Yatch, Council Diplomatic Shuttle, Gnosis, Sunesis, Echelon, Primae, Apotheosis,âŚ) that are unique in my hangar and just take dust.
Because I know that if I fly that I will get ganked immediately and lost something rare that cannot be replaced.
Why donât you make BPO that are not tradable and produce ships that are not tradable either?
That way player would have a higher incentive to perform the tasks AND fly the result of their work instead of letting it take the dust.
I tried several minutes with maximum zoom, but the positioning precision for transits just does not allow you to match all ⌠this is the best I could get.
Was there an live update? Currently PD is bugged, when you finish a sample all points for the level are counted not only those you just got, long animation. When you start a new sample it flickers a couple of seconds before you can start ⌠happens since ~10min.