Talk:Sparkline in unicode: Difference between revisions

:::::: Ah ... You though I was defending particular drafts ? No, several things need fixing.

:::::: My point is that while checking and fixing these drafts, we '''can't''' reliably use your 2nd test (0..7999) and claim that there is a single correct output which it should produce.

> if the task is "'''assign 8n items to 8 bins as equitably as possible'''" ...

:: That '''isn't''' the task ! :-) Those words are '''yours''' – they are not from the task description at all ...

:: It would be perfectly reasonable and understandable for you to aim for that goal, but it '''isn't''' specified by the task, and, as it happens, it actually leads to logical contradictions, and requires a fudge, if taken literally.

:: Many (probably most) charts leave some margin to left and right, allowing for a scale of possible values that is larger than the range of observed values. Perfectly resonable, and it doesn't require all 8 buckets to be equally full.

:: Mathematica and R are graphing the data on the scale 0..8000. That's at least as reasonable as your approach (mathematically a little more simple and coherent too, as it happens), and it would be just as sensible (possibly even more) to follow their lead as yours. Both approaches are fine.

::: The '''point''' is that your presentation of the 0..7999 test as a reliable diagnostic instrument which 'should' yield a unique pattern with only 3 levels is '''misleading'''. It can produce false negatives, and doubtless false positives too. The Mathematica and R approach, equally correctly produces '''5''' of these low resolution levels with that data, and its deliberate clustering around margins.

::: What you are suggesting is cetainly one way, but '''not''' the only way.

::: The task is '''not''' to fill buckets evenly. The task is to produce a vertically proportional sparkline. Your test does not have a unique correct output. No reason that it should, but if we misrepresent it, we will mislead people [[User:Hout|Hout]] ([[User talk:Hout|talk]]) 07:14, 27 February 2019 (UTC)