ISC Quarterly Update for the R Consortium (submitted 2019-03-28)

[imkemayer]

Thank you for participating in the R Consortium ISC Funded Projects program. Please complete this form to provide your quarterly update on your project progress, concerns, and issues needing ISC feedback. This feedback is required each quarter so that reimbursements can be processed.
R Consortium will use this feedback to share project progress with ISC and Board members, as well as general updates on the R Consortium website and social platforms.

Please detail milestones since the last project report, including any success metrics ( e.g. number of downloads, attendees at recent event, etc)

• Official release/publication of the R-miss-tastic website (https://rmisstastic.netlify.com),

• Frequent feedback from R users to our CTV and website on adding new R-packages to the CTV and developing workflows on the website,

• Work session with new collaborators during DAGStat conference in Munich (Germany),

• Advertising of the website during various conferences/events (in France, Germany, USA, Australia), in blog posts and on twitter.

Provide a public facing update on the project *
This will be published on the R Consortium website and blog

In January we have officially released our website https://rmisstastic.netlify.com/ that houses articles, tutorials, data sets and a first set of workflows around a small set of popular R packages. We will continue to provide more tutorials, assistance in choosing and using existing R packages, and data sets as time goes on. To help make this project as robust as possible, we encourage authors to submit their articles or works, and reviewers to review their placement on the platform/website, either by contacting us via our website or by submitting changes directly in our github repository https://github.com/R-miss-tastic/website.

We recently held a collaborators meeting in Munich during a statistics conference to meet in person with motivated contributors who answered our call for joining the project, Steffen Moritz and Dominik de Sordi. In future meetings during the next quartal we will tackle together the final step of our proposal, future extensions and beyond: we aim at creating a benchmark of existing methods for different kinds of data (both synthetic and real data), missing values mechanisms, tasks to be fulfilled, etc. An important aspect is that this work should allow other researches and data scientists to re-use/copy our R code to compare their own method to a maximum of existing methods without having to re-implement the comparisons every time themselves. This will allow for better comparability and credibility of newly proposed methods. Finally such a unified benchmark has not been proposed anywhere else yet.

By having a community involved in this, we can then have useful discussion on the benchmarks and approaches to multiple imputation, even organize challenges to find the best imputation methods, perhaps in a similar fashion to the M4 forecasting competition.