VA Project Management Protocols

The following protocols are to create standard system for managing data, scripts, documentation, etc. for VA projects on res10.

Folder Structures

  • In general, folders have the hierarchical structure Project - PI > SubProject (optional) > Data/Tasks - Primary user.
  • Top level folders should be named with the convention Project Name - Project PI. For example, the main folder for the DEcIDE project will be DEcIDE AHRQ DM - Roumie.
  • Subfolders used to accomplish specific tasks should be named with the convention Task Name - Primary User. For example, the subfolder where Sandeep plans to run a series of data checks could be called Out Pt Meds Data Checks - Sandeep. Optionally, the user may opt to add a date to distinguish similar folders, e.g. Out Pt Meds Data Checks - Sandeep - 2009-03 and Out Pt Meds Data Checks - Sandeep - 2009-07. Note dates should be written to ensure alphabetical order = chronological order.
  • Almost all contents related to a given task should be in the same folder, e.g. scripts, output, documentation relating to that task, etc. To save space, big datasets that don't need to be modified don't have to be copied into the folder, provided the dataset is read only and can't be accidentally modified.
  • Data folders. Some folders do not have a task associated with them, for example a data folder may simply hold data nothing else. Data folders should be labeled to indicate if the data is "raw", "clean", or "analyzable". For example, the output from an initial data pull may go into Out Pt Meds - raw - 2009-03. The folder Out Pt Med - clean - 2009-05 may contain the cleaned data from the 03 pull as well as the cleaning script and relevant documentation. Analyzable data is often used just for specific analyses, so it may be best placed in a specific task subfolder.
  • In general, we've found that folders work best with a singer user controlling all they do in that folder. However, having a main folder like Robert with a bunch of subfolders does not work well. Break up your work into task specific folders and identify them as yours using the convention above, e.g. Create Raw Persistence Windows - Robert. * All work for a project should be contained somewhere within that project's main folder.

User Specific Preferences

  • Users will generally have their work focused in specific subfolders.
  • Users may use any sensible file naming convention they wish within their own task specific subfolders.
  • Users should have a good system of documentation and version control kept within their task specific subfolders.
  • Programming Tips For Statisticians for suggestions and more information.
Topic revision: r1 - 07 Apr 2009, RobertGreevy
 

This site is powered by FoswikiCopyright © 2013-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback