Investigation discrepancy between Jupyterlab - Jupyter notebook for R script / Merge revisions

[eliselavy]

I have attached two pictures of which words are assigned to the topics by LDA in the journal webpage and how they appear in JupyterLab on my machine. Since I added the labels manually, only the JupyterLab version makes sense. After that, all the other results are jumbled. I have scouted the internet to look what might have provoked the different results, but I have not found an answer in the queries. I notice that the error is consistent, not random. The words produced in the journal webpage are the same that come out when I run the program in Jupyter Notebook, while running the program in RStudio and JupyterLab produce a different set of results. So I guess there must be a hidden parameter somewhere.

I also changed the "SI_metadata.csv" file to respond to the comments of the reviewer, so please include the new file. I have also changed runtime.txt

TOPICS in the journal:

TOPICS in JupyterLab

[eliselavy]

Problem occurs also for the following notebook: https://github.com/jdh-observer/jdh001-9HcfToh7EYm8

Any version of R Studio that supports R 3.5.0 (april 2018) or newer has to work according to my information (and experience). The R version initially used to create the scripts was R 3.5.0 as well, but the scripts also run on newer versions.

[eliselavy]

Kernel not found with FROM rocker/binder:4.1.1

Get back to version FROM rocker/binder:3.6.3

[eliselavy]

On mybinder the good topic modelling

[eliselavy]

with image FROM rocker/binder:3.6.3 OK with image FROM rocker/binder:4.1.1 NOK

[eliselavy]

Locally 3.6.3: My binder:

But not the good result

=> solution download the notebook from mybinder

[eliselavy]

for the following notebook not linked to the R's version but more to the version of the file: https://github.com/jdh-observer/jdh001-9HcfToh7EYm8/blob/main/stuff.RDS