Idea: Re-write the workshop to tell a story

[gvegayon]

Right now, the workshop has somewhat disconnected parts that walk through the different features of netdiffuseR. Instead, we could take a dataset as a baseline and write the workshop around that. I envision something like the following:

1. Intro to diff of innovations theory.
2. Read diffusion data in.
3. Inspect it: visualizations and basic tables.
4. Diffreg model.
5. Simulations.

This would allow building something more interesting for the attendees who would fit great with their code. We can use all the materials in the workshop to add them as vignettes for the package and a baseline for the workshop contents.

attn @aoliveram @twvalente

[gvegayon]

Decisions:

We will use KFP across the examples. We will run the workshop as if we were writing a paper using netdiffuseR.

• (skip) Move the getting started to misc.
• Start with Diffusion Theory.
• (skip) Move "The netdiffuseR package".
• Read in data: Either your own data or KFP. Focusing on the dynamic survey: survey_to_diffnet().
• Visualization: Focus on plot_adopters(), plot_threshold(), and plot_diffnet(). For the problem, look at the "I trust my friends" example.
• Statistical analyses:
  • Moran's I (via the summary function).
  • Lagged regression model (diffreg).
  • For the problem, we could ask them to add some variable to the dataset (like in a logit) and ask to report the results.
• Simulation:
  • rdiffnet_multiple using the same network we have been working with. Two examples, changing threshold/seeding strategy.
  • Example using disadoption function. Exercise: Create a disadoption function that makes younger folks disadopt sooner.

[aoliveram]

Okay, this is what I had been doing in the branch issue-2-tell-a-history:

1. I have added more details about the workshop to the 'Home' section and included links to refer to older versions of the workshop and the project repository. Also, a Miscellaneous section was created, with 'Getting started', the 'intro to netdiffuseR', and the 'classic diffnet analysis'. The 'intro to netdiffuseR' was the only subsection I touched a little bit.
2. I moved 'Diffusion Theory' to the start. Only stylish modifications.
3. Some chunks were removed from 'Reading Data'. The function to be used is indicated in the subtitles, and some emphasis was added in survey_to_diffnet() using the kfamily dataset. Basically an adaptation from 'classical diffnet analysis', so it still needs to be improved. No further modifications.
4. In 'Visualization', I added only a chunk to create kfamily_diffnet_21, a diffnet object from subset(kfamily, village %in% c(21)), and run the main plotting functions: plot_adopters(), plot_threshold(), and plot_diffnet().
5. The 'Statistical inference' section was reorganized slightly for clarity, but there is no mention of kfamily yet. Also, could you review the Appendix (¿?).
6. In 'Simulations,' I added the Network threshold equation and reorganized the chunks to make them more intuitive. Also, I fixed some errors. For example, diffnet_disease and diffnet_rumor had opposite names.
7. There is 'Simulations pt.2', basically because I think it would be too long as a single section with all the simulation chunks, even if I removed all non-essential ones. It uses kfamily_diffnet_21 a lot, so it points towards the goal. Maybe Example 2 could be the exercise?

As you can see, the basic structure is done, but there is room for improvement.

@gvegayon @twvalente what do you think?