Day 0 is to think about how to do it, but the answer is pretty simple: use a version control system, like Git. Because it tracks every bit of what you do, allows for easy back ups, and makes it easy to continue working on a different machine in case you forget to take your laptop adapter home :)
- Day 1: keep an electronic lab notebook (e.g. a version control system; read Git from the Bottom Up)
- Day 2: carefully select data you build on (can you indeed share it with the rest of your arguments in your next paper?)
- Day 3: do you research and store everything
- Day 4: integrate data repositories in your data analyses, e.g. rrdf and knitr
- Day 5: if you like scientific dissemination, collaboration, and progressing science, share your data in public repository, like FigShare, Data Dryad, Dutch Dataverse, 3TU.Datacentrum, DANS, etc. (that's a lot of D-D-D-Data...) or in a domain specific database, like WikiPathways, XMetDb, or DrugMet. And data copyright and licenses and particularly, whatever you chose, be explicit about it and don't let others guess (wrong).
- Day 6: think ahead of reuse, and suitable formats. Consider semantic web and linked data.
- Day 7: did you get impact? Think DataCite, ImpactStory, and Altmetric (and ORCID and DOI along the way).
And here are the slides: