RFC: Statistics in Subsurface

[Christof Arnosti]

Hi Dirk,

Thanks for your response.

Am 14.05.20 um 18:37 schrieb Dirk Hohndel:
> Hi Chrisof
>
>> On May 14, 2020, at 3:30 AM, Christof Arnosti <charno at charno.ch
>> <mailto:charno at charno.ch>> wrote:
>>
>> Let me give some thoughts from a more-or-less outsider perspective in
>> this discussion.
>>
>> From a UI-Perspective, I would prefer the layout to be "Dive List top
>> left", "Filter top right", "Stats selection bottom left", "Stats
>> display bottom right". This is with the reasoning that (in the
>> western world) we work from top left, top right, bottom left, bottom
>> right, and the logical Workflow would be to have the dive list as
>> input, which is then filtered, then the statistics selected and
>> finally the output. This workflow could also be applied as sort of a
>> wizard for mobile devices.
>>
>
> In insulation that may be true. But our existing UI is not up for
> discussion. And that very intentionally has the information that the
> user interacts with (information tabs and profile) on top, and the
> selection of dives on the bottom. Switching that around to show
> statistics would lead to a horrible user experience.
I see. Makes sense to be consistent here.
>
> On mobile these would have to be different pages, anyway, so there the
> 'layout' question is fairly moot
Agree.
>
>> For selecting the graph type (the bar vs boxplot discussion): Could
>> this also be implemented as an option in the "Stats selection"
>> section? I think it's obvious that some people prefer the more
>> advanced boxplot version, and some the easier to understand bar/line
>> graph version.
>>
>
> Every time we opt for "oh, let's make this selectable" we
> significantly increase the amount of code that needs to be written and
> tested, and we make the UI more complicated by adding more options.
> We already have way, way, way too many options. And we constantly find
> that yet some other feature has bit-rotted and doesn't work anymore.
> Or that some changes to our code break something else that doesn't
> have an active developer anymore.
>
> Today Subsurface is de facto maintained by about five people, three of
> which contribute in very narrow slices that are "theirs", and the
> other two (Berthold and I) try to keep everything else working.
Makes sense. Bitrot is a pain I just know too good (I'm currently
procastrinating to fix some legacy C++ code on a microcontroller that
won't cooperate with updated host software... The horrors!).
>
>> Another (third? fourth?) option I just tought of was a boxplot with
>> included histogram, where the histogram is displayed as color
>> (instead of a curve). I have attached an image of a short mockup
>> (Where red means more, and green means less). I'm not sure if that's
>> a good idea, but at least it's an idea ;-)
>>
>
> It's definitely an idea. It's geeky and cool. I don't think it will
> help accessibility (in the sense of being easy to understand for the
> casual user).
I agree, if there is no selectable graph style this should NOT be the one.
>
>> About graph orientation: I strongly agree that the bars (or boxes or
>> whatever) should be vertical, so that the time-axis (or trip axis or
>> whatever) is the z axis.
>>
>
> You mean time / categories should be the x axis, correct?
Yes, exactly. Sorry for confusing X/Z. For me it's just way more natural
to process data from left to right instead of top to bottom, especially
if it's time-based (which I think it will be in the most cases -
categorizing by trips will probably also be sorted by time?).
>
>> This is (at least for me) the natural way to read graphs, and also
>> what's currently done in the dive-display.
>>
>
> Which dive display? The dive list has time as the vertical axis. Do
> you mean the dive profile? That's a single dive, not a collection of
> dives. Very different.
Yes, the dive profile. I know it's a different thing, but still it's a
graph and time is on the x axis. I think this would add some additional
confusion if there are differently-oriented graphs.
>
>> And as a last point a proposal for a little visual gimmick: I would
>> really like to have the value-axis for depth turned around, so that
>> the 0-point is on top of the graph (like in the current dive-graph).
>> With this, the visual representation of the data is the same as in
>> the physical reality, lower means lower.
>>
>
> Yes, that makes perfect sense for a dive plot (which is why we do it
> that way). For statistics I would find it absolutely painful.
I wouldn't ;-) But as said, it's just a gimmick.
>
> /D 
Best regards
Christof
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.subsurface-divelog.org/pipermail/subsurface/attachments/20200515/d58affc8/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kijinokeijnnhkpj.png
Type: image/png
Size: 1581 bytes
Desc: not available
URL: <http://lists.subsurface-divelog.org/pipermail/subsurface/attachments/20200515/d58affc8/attachment.png>