# Milliways: Taming Multiverses through Principled Exploration of Data Analysis Paths
## Overview
This repository contains the supplementary materials for the submission `Milliways: Taming Multiverses through Principled Exploration of Data Analysis Paths`.
The repository is organised as follows:
- `milliways-video-demo.mp4`: this is a video demonstration of how Milliways can be used to performed a principled exploration and evaluation of multiverse analysis
- `milliways/`: directory that contains the implementation of Milliways in JavaScript and Svelte.js (see below for more details)
- `milliwaysR_0.1.0.tar.gz`: compressed file which contains the implementation of the R interface to Milliways
- `example/`: contains the Rscripts used to generate the hypothetical dataset used in the example, as well as the corresponding multiverse analysis
- `example-boba/`: contains the same analysis (as previous) implemented in boba. We provide the files which are necessary to load the results using boba
- `user-study/`: contains materials for evaluative study, which includes:
- training interface (template.html contains the visualisation tool, analysis_doc.html contains the EMAR)
- task interface (template.html contains the visualisation tool, analysis_doc.html contains the EMAR)
- interview protocol
- participant details
- transcripts
- `revision-letter.pdf`: Revision letter containing reviewer comments (from prior submission to VIS 2023) and how they were addressed
Note #1: while the template.html files can be accessed directly in your local browser, it is better to run them on a local server: `python3 -m http.server`
Note #2: We redesigned the UI of Milliways after the user interviews. We include an updated `template-new.html` file which contains a standalone example of Milliways with the updated UI.
## Installation
### Installation of the Milliways Visualisation tool
To install `multiverse-vis` using `npm` from your local repository you need to call:
```
cd milliways
npm install
```
To watch for changes and continually rebuild the package (this is useful if you're using npm link to test out changes in a project locally):
```
npm run dev
```
### Sharing
To create a static HTML page of the interactive multiverse visualisation tool, with the assets pre-loaded users can call:
```
npm run compile
```
This will create a `template.html` file, which will work as a standalone HTML document.
### Installation of the R Interface to Milliways
Unzip the directory and open the .RProj file `milliways.RProj`
Note #3: the following code should be executed in R
You will first need to install devtools:
```
install.packages("devtools")
```
Then into the console, execute:
```
devtools::install()
library(milliways) # note that using devtools::load_all(".") will not work
```
After you've installed the R library, you will need to install Milliways as an npm package:
```
milliways_install()
```
To open Milliways with the results of a multiverse analysis, you can use the `start()` function. This function requires four arguments:
- `results`: (path to) a JSON file which contains the estimates from each universe in the multiverse analysis.
- `code`: (path to) a JSON file which contains the code used to generate the multiverse object.
- `data`: (path to) a JSON file which contains the dataset used in the multiverse analysis.
- `analysis`: (path to) the Explorable Multiverse Analysis Report (EMAR) HTML file
We provide samples of this in the directory `dist`. For example, to initialise Milliways with the multiverse analysis of the hurricane dataset, you can call the following command:
```
start(
"dist/hurricane/hurricane-data.json",
"dist/hurricane/hurricane-code.json",
"dist/hurricane/hurricane.json",
"dist/hurricane/analysis.html"
)
```
