more_info.md

In ncsa/DataHarmonizationPipeline: An automated statistical analysis pipeline in R for use on public health datasets

Further Information: Design, Implementation and Documentation

Code Design

• As mentioned in the README, the package is the implementation of a proof-of-concept statistical pipeline to analyze public health datasets, developed previously at NCSA.
• The main objective of code design is modularity: we want the user to be able to choose which steps of the pipeline they would like to execute.
• The pipeline involves three main tasks: Data Preprocessing, Data Separation and Analysis. Consequently, we have three R scripts, pertaining to each task.

Code Implementation

• There are 4 R scripts: DataPipelineDriver_New.R, DataPreprocessing_New.R, DataSeparation_New.R and DataAnalysis_New.R.
• DataPipelineDriver_New.R - contains a main driver routine that runs all functions sequentially according to the pipeline.
• DataPreprocessing_New.R - implements the pipeline blocks that preprocess the data.
• DataSeparation_New.R - implements the pipeline blocks that separate the original dataframe into discrete and continuous data frames.
• DataAnalysis_New.R - implements the pipeline block that analyzes the data (performs PCA, etc.)

Code Documentation

• Documentation of code in the package has been done using roxygen2.
• Simply type in ?function_name in the RStudio terminal to bring up a pop-up listing a function description, argument descriptions, usage example and return data type and description.
• roxygen2 is used in our code because of
• Standard usage in most CRAN packages
• Easy maintainability
• Automatic generation of nicely formatted R Markdown documentation files

ncsa/DataHarmonizationPipeline documentation built on May 30, 2019, 2:05 p.m.