Proposed FOSS4G talk (abstract): Interfacing QGIS processing algorithms from R

[florisvdh]

The idea rose to give a talk at FOSS4G (26 June - 2 July, Prizren, Kosovo) and perhaps at some other occasions too.

Below is a first draft of an abstract, available for further editing in a google document (if making suggestions there, please do so using a google account, not anonymously). The deadline is 19 Feb, so this should be finished by 18 Feb the latest. EDIT: deadline postponed to 28 Feb.

Interfacing QGIS processing algorithms from R

R is widely known for its unsurpassed availability of well documented statistical functions and packages. Moreover, the R ecosystem has brilliant packages for geospatial work, for example sf, terra, … and various packages that build upon these. On the other hand, a dedicated Geographic Information System such as QGIS offers specific processing algorithms that are either not available in R, or don’t necessarily achieve the same level of performance in R. However it is possible to combine the strengths of R and QGIS. We discuss the current state of R packages that interface processing algorithms provided through QGIS. We demonstrate the usage of the most important functions with some examples. Finally, we shed light on future development directions and hopefully get some feedback from the audience.

Pinging @paleolimbot, @JanCaha (who already responded to the first ideas by email)
And pinging @jannes-m @Robinlovelace (please see also my message to you on Discord!)

[jannes-m]

Great initiative +1, I have made a few suggestions in the Google doc which you might find helpful.

[paleolimbot]

Thanks for leading this (and development in this repo for some time!).

I'm happy to give feedback on further iterations of this - I think that the abstract is good and that presenting this benefits R and r-spatial in its demonstration of R as a "glue" language.

[florisvdh]

The google doc now also includes a tentative author list + affiliations to submit with the abstract. Suggestions to make changes are welcome; please complete the affiliations!

Submission will also need to assign a number between 1 and 4 indicating the level of technical complexity of your contribution; I suggest to choose level 2:

Any idea about Give indication of resources (video, web pages, papers, etc.) to read in advance, that will help get up to speed on advanced topics.? I think it's best to state 'not applicable', unless you think we should consider this an 'advanced topic' and perhaps refer to the README (should be updated by then), the repo, or the original discussion at rspatial/discuss.

[florisvdh]

Good to know: today, the submission deadline has been extended up to 28 Feb!

However I suggest to finalize this weekend, or beginning next week by the latest.

https://twitter.com/foss4g/status/1626536948184780802

[florisvdh]

@paleolimbot @jannes-m please can you add your correct affiliations in the gdoc? Thanks!

[jannes-m]

@florisvdh thanks for putting me on the author list but so far my role regarding qgisprocess has been, if any, only an advisory one. Therefore, it's really OK if I am not one of the co-authors. If you want to put me there nonetheless, my affiliation would be: GfK Geomarketing GmbH, Sophie-Germain Str. 3-5, 90443 Nuremberg, Germany

[florisvdh]

Hi @jannes-m, thanks for your response – I understand. Nonetheless I'd expect that you will be package co-author / contributor at the time of FOSS4G, at least because of the documentation part, given your pending but important vignette contribution in PR #31 and expecting that we can make it to CRAN before FOSS4G. For me that would be enough reason to add you to the presentation's author list for FOSS4G. Would you agree?

Anyhow, we can also decide on your presentation co-authorship later and not pass your name now, if you prefer.

[jannes-m]

Right, the vignette, I should give it a go again. OK, after having taken care of geocompx/geocompr#911, I will look again into the vignette.

[Robinlovelace]

Hi all, just picking up on this finally, will make some suggestions in the doc if that would be helpful?

[florisvdh]

Sure @Robinlovelace, thanks!!

[florisvdh]

I may submit tonight (after 16:00 UTC that is); if you planned to re-read, now is a good time. Please hold me if you need some more time!

[florisvdh]

Done!

Review link: https://talks.osgeo.org/foss4g-2023/talk/review/E7ZQADYWVKFE7GDA3EKAUJF7CFHHS7JW

[jannes-m]

This really reads well, great job! The only thing I stumbled across was "(batteries included!)", I have no idea what this should mean.

[florisvdh]

Thanks Jannes; then it needs to be reworded. 'Batteries included' should refer to the fact that statistics are included in the default installation.

@Robinlovelace I had changed your original in the default installation, with ‘batteries included’ to: in the default R installation (batteries included!)

Best propose an update in the google doc, then I'll update the FOSS4G proposal (remains possible to do that before the deadline of Feb 28).

My suggestion is to drop the batteries.

[Robinlovelace]

This really reads well, great job! The only thing I stumbled across was "(batteries included!)", I have no idea what this should mean.

Means that with R you get going quickly with data analysis, without having to go through additional steps. The metaphor: devices that lack batteries cannot be used out-of-the box because you have to find the right batteries, buy them, put them in. In some languages you can do lots but set-up is a pain. With R you do install.packages("sf") and bingo you can do perhaps 50%+ of common geographic data operations, more with install.packages("terra") etc.

[jannes-m]

Thanks for the clarification! Ok, if this is a well-known computer science metaphor I just wasn't aware of, leave it in, if not, leave it out, would be my suggestion.

[florisvdh]

I'm in favour of dropping it to avoid confusion and ease readability, since the meaning is already there without adding it.

[Robinlovelace]

+1 to readability, it's not a well-known in computer science or any area to my knowledge so in hindsight may not have been best to include it. I think I may have written the phrase in the comments and that found its way into the main text. Good to make things clear, I'm a fan of plain English, and it's better without it.

[florisvdh]

I think I may have written the phrase in the comments

Right, I remember!

It's done, thanks for the feedback. https://talks.osgeo.org/foss4g-2023/talk/review/E7ZQADYWVKFE7GDA3EKAUJF7CFHHS7JW

[florisvdh]

Just dropping by to say that the FOSS4G talk has been accepted 🙂. So I'll register and go give the talk.

[florisvdh]

A first draft of the slides has been made. You can leave comments or add pull requests (e.g. improving layout) in https://github.com/florisvdh/foss4g-2023-qgisprocess!

Still planned:

• add a qgis_search_algorithm() function and use that in the slides
• improve layout

[JanCaha]

Looks good. Might be slightly longer, but i dont know what the time limit is.

[florisvdh]

Thanks for having a look @JanCaha. The talk must remain within 20 minutes. Yesterday I did a try-out in front of colleagues. It was found that the presentation should definitely not be longer, it takes some time to explain things. I will process other feedback that I've got in an updated version and keep you posted; probably from the presentation repo itself where we can continue discussion.

[florisvdh]

Further discussion on the presentation is continued at florisvdh/foss4g-2023-qgisprocess#1.

Feel free to join and make improvements, by June 23 the latest please.