? ppl attended
Open exchange
- new paper by Clemens:
- https://github.com/nevrome/mobest.analysis.2022: uses ancient genomic data to reconstruct where people may have come from
- Clemens tried to make it reproducible, but that’s difficult with high computation set ups, some of the submission scripts will probably be re-written for other set ups
- is a meta analysis, therefore no problem with big data accessibility –> only links to other data sets
- data is shared by a lab in Harvard, so hoping the data will be available for the foreseeable future (copying the data may be difficult for legal reasons)
- maybe the data will be later made accessibly more openly (cooperation coming up)
- all analysis is written in R, some Haskell is for management of computational pipeline (Haskell shake, similar to R targets), and some more command line tools
- changing APIs can be a problem for reproducibility
- Thomas Huet: tries to set up a Workflow with google maps and asks for recommendation
- extract spatial data from remote sensing, to get a learning data set for machine learning
- R gee package or ee by Python?
- Python is used by more people
- Python is especially useful for spatial analysis (most GIS is written in Python), R would work also, though
- Pyhton environments is more hard to set up
- machine learning part is probably also written in Python so maybe better stay with one language
- deep learning workshop at CAA is booked out
- there was an R-workshop in Pisa (Francesco Carrer and Joe Roe): good feedback, probably will take place next year again (this was the fourth time already)
- if you have students that need an R intro
- archaeovis-package by Sebastien Plutniak: https://github.com/sebastien-plutniak/archeoviz
- mostly for x-y-z projection of data and showing the stratigraphical position of artefacts after excavation
- tries to make non-scripting people use shiny
- preprint is submitted: https://zenodo.org/record/7674699
- is at PCI for review
- more GUIs for R are being developed
- in Clemens part of the workshop in Australia: 20-25 ppl attending
- quite interesting talks, quite theoretical in part, very diverse topics, but always linked to quantitative methods and R scripting in these domains
- but Clemens just saw a couple of talks
- software reviews: https://research-squirrel-engineers.github.io/Impuls_SoftwareRezensionen_DGUF/Draft_EN.html
- “digital reviews” in archaeological advancements
- PCI offers the possbility to review software
- peer reviewing software is something else as reviewing software
- archeoFOSS community: already engaged in this topic?
- should the SIG engage?
- if there would be a centralised platform to find other people’s experiences it would be sensible
- a list of recommendations / reviews might be already useful
- “lists” of software are already avaiable
- software is changing all the time: so reviewing would be one step, but it should be starting point of involvement, improvement – make a open source collaboration
- if done correctly this could be a great way to improve software (otherwise it might be very annoying for the author to be criticesed and not helped to improve)
- another point might be: making a software more known (also positive for the author)
- review paper: a paper that says “if you want to do this kindof analysis, you have these options…”, like a survey
teaching materials: current state of affairs
- https://sslarch.github.io/MapofComputationalArchaeology/
- new: introductory information added
- looks great!
- feature to maybe add: click on the tag and filter for the tag
- put it on our homepage (link or embed?
- where on the website
- under “research output”?
- or rather on top as its own tab?
- maybe important links on a side board (but check mobile view: it will go down to the botton)
- let’s discuss this on the github: issues for the website (also, should you find a problem with the map)
- topic for next meeting: how to improve our website
Petr Pajdla on Quarto
- doesn’t replace Rmarkdown (development will be continued) and written text stays markdown
- Rmd documents should work with Quarto
- Quarto is a Command Line Interface, more generic than the R dependent Rmarkdown
- very well written introduction: quarto.org
- Quarto is build on Pandoc (may be needed to be installed seperately, esp. pandoc-citeproc for citation)
- Quarto is now shipped with Rstudio (you may be able to use it out of the box)
- projects workflow:
- quarto documents (extension qmd)
- has a similar YAML-header, but can also be seperated as _quarto.yml
- splitting one project into several documents easier than with bookdown
- integrates author, name, email, orcid: easier to include than with Rmarkdown
- publishing:
- quarto pub hosts slides, etc whatever, every easy to use
- github pages easy to set up
- figure captions: now not any more put next to the ```{r
- this was a bit unwieldy in Rmarkdown if the labels and captions got very long
- in Quarto same information as a comment #| –> better readable
- labels need to be fig.xxx for figures and for tables tab.yyy
- very easy to have a qmd document to become a presentation
- ::::{.colums} creates a div environment for tables for html
- revealjs renders to html
- same code goes to same output in html, PDF and docx <- details needs to be checked for the future
- css might not be well translated into other outputs, basic text colours will be the same
- citations work the same as in Rmarkdown (put the bib file and csl in the YAML)
- works also in visual mode
- integrating ready made images works the same -> {#label}
- label: “#fig-xxx”
- reference:"@fig-xxx"
- check cross-references
- supports other languages?
- Python, R, Julia are implemented - can be used in the same document
- can the community provide kernels to enlargen the pool of supported languages?
- knitr or jupyter can be chosen for rendering, thereby using jupyters kernels
- more languages are supposed to come, if sth is installed on you computer, it should be possible to use
- Rshiny in Rmarkdown was a great interactive possibility in the documents
- can be done in Quarto as well
- example of Quarto used for a course website: https://gia.joeroe.io/ (source at https://github.com/joeroe/gis_in_archaeology)
- worked really well
- Petrs Diss uses Quarto: https://github.com/petrpajdla/dataInfrastructures
Advertisement by Francesco Carrer
- 2 fully funded 1-year MRes (master of research) + 3-year PhD for UK students (or foreign students that obtained their degree in the UK) at Newcastle University.
- (1) Spatial Analysis in Archaeology: “Spatial Analysis and Modelling to Investigate the Historic Character of Landscapes” - supervisor Francesco Carrer, co-supervisor Sam Turner, industrial partner Historic England; knowledge of GIS essential, knowledge of R or Python desirable.
- (2) Mathematical Archaeology: “Spatiotemporal modelling of Alpine agricultural revolution” - supervisor: Andrew Baggaley (Maths), co-supervisor Francesco Carrer, industrial partner Dolomiti National Park; experience with techniques for finding both numerical and analytic solutions to differential equations essential, working knowledge of Python or Fortran.
- For more information, get in touch (francesco.carrer@newcastle.ac.uk) and check the website (https://geospatialcdt.ac.uk/studentship-projects/)
next we will meet at the CAA, more info (where and when) will be distributed on Slack
- to be discussed:
- Mastodon and Matrix
- how to improve on our websit
- May: talk on morphometrics
- June: Zach offers to introduce LaTeX