Communicating Results¶
People are busy, hence yo ushould breakdown results into different levels of detail
Research Paper¶
- Title
- Author list
- Abstract
- Body
- Results
- Supplementary Materials (details)
- Code/Data (more details)
Email¶
- Subject
- Summarize findings in one sentence
- Email body
- Brief description of problem
- Recall what was proposed
- Recall what was executed
- Summary findings in 1-2 paragraphs
- If you want to suggest future actions, make them concrete
- If you have questions for them to address, make them yes/no
- Attachments
- Literate Programming File
- Links to supplementary materials
- Code/Software/Data
- Github Repo/Project Website
RPubs¶
Free site to publish work
Reproducible Research Checklist¶
โ Start with good science
- Garbage in, Garbage out
- Coherent, focused question
- working with good collaborators
- Something thatโs interesting to you
โ Donโt use GUI softwares for analysis, as it is hard to reproduce
โ Donโt do things by hand
- Using spreadsheets to perform operations
- Editing tables/figures manually
โ Document unavoidable manual operations (not as easy at is it sounds)
- Downloading data from a website
- Moving data to outside project folder
- Splitting/reformatting data files
โ Teach a computer to do tasks
- Downloading data
- Web scraping
โ Use version control (such as Git)
- Helps slow down and perform tasks step by step
- Add changes in small chunks
โ Keep track of software environment
- Write cross-platform code
- For eg, in python be careful when working with paths. Use
os.path.join(folder, file)
instead offolder + "/" + file
- Computer Architecture
- CPU, GPU specs
- Operating System
- Software toolchain
- Compilers/Interpreter
- Programming languages
- Command shell
- Database backends
- Data analysis softwares
- Supporting software
- Libraries
- Packages
- Dependencies
- External dependencies
- Websites
- Data/Software repositories
- Remote databases
- Version Numbers
โ Donโt save output independent of code
โ Set your seed for random number generator