search

metarelate / statistical-process

BSD 3-Clause "New" or "Revised" License
0 stars 1 forks source link

form of difference #14

Open marqh opened 6 years ago

marqh commented 6 years ago

In the current Improver code, this concept is proposed to be encoded as:

spp__form_of_difference = spv__forward_difference
<>
        a               rdf:Property ;
        rdfs:label      "" ;
    dct:description "" ;
        dct:identifier  "" .

Is the identifier form_of_difference a clear and long term identifier?

What 'description' would we give this concept?

The first known value for use with form_of_difference is forward_difference

<>
        a               skos:Concept ;
        rdfs:label      "" ;
    dct:description "" ;
        dct:identifier  "" .
cgsandford commented 6 years ago

Can I suggest possible values of the form eg "spv__forward_increment"? "Difference" could refer to a variety of things, such as ratios - don't know whether we want to capture that in the attribute name or restrict the definition more.

MarkWorsfold commented 6 years ago

Hi, I think its a good idea to avoid using the term "difference" - like Caroline said - it can have so many different meanings.

Me and Gavin have been discussing this and we think that a modified version of Carolines proposal would be: spv_increment_direction = "forward/backward"

For example if an identifier states: spv_increment_direction = "forward" We would mean that we are incrementing from left to right of the array. But we need to do something in the documentation to make this absolutely clear.

This can be coupled with another piece of metadata where we state that the stepsize is N cells. spv_increment_step = N

marqh commented 6 years ago

Hello all

Looking from the outside, I can see how spp_increment_direction is a more widely applicable term.

An option for defining direction, chiming with forward/backwards, is to link it to the coordinate values. So the increment direction is either with increasing coordinate values or decreasing coordinate values. I might use terms like coordinate_positive: the direction of increasing coordinate values and coordinate_negative: the direction of decreasing coordinate values

A different option is to link direction to the indexing scheme, so the increment direction is either with increasing index or with decreasing index.
I would interpret forward/backward as being an indexing direction, as described by @MarkWorsfold

Is it preferred to identify direction via indexing or via coordinate metadata?