R Utility Toolbox

A collection of my random but frequently used R functions to make life easier as a statistician and data scientist

Overview of Toolbox

• Data Cleaning/Preprocessing
  • functions for basic data preprocessing and cleaning
• Modeling
  • functions for fitting and evaluating prediction models with consistent inputs and outputs
• Plots and Tables
  • functions to create custom ggplot and kable themes as well as to provide convenient wrappers for commonly used plots and tables
• Rmarkdown Templates
  • Rmarkdown templates for generating reports from simulation studies
• Miscellaneous

Data Cleaning/Preprocessing

• clean_functions.R: functions for basic data preprocessing and cleaning

  • basicDataSummary(): function to summarize basic data oddities (e.g., NAs, constant columns, missing factor levels)
  • removeNACols(): function to remove all columns with at least one NA value
  • removeConstantCols(): function to remove all columns that are constant

• filter_features.R: functions for filtering number of features in high-dimensional data

  • filterByAssoc(): function to reduce number of features by keeping those with the highest univariate association with the repsonse (measure via ANOVA)
  • filterByCor(): function to reduce number of features by keeping those with the highest correlation with the response
  • filterByCorBootstrap(): function to reduce number of features by keeping those with the highest correlation with the response, averaged across multiple bootstrap samples
  • filterByLasso(): function to reduce number of features by keeping those with the largest magnitude coefficients from Lasso
  • filterByRF(): function to reduce number of features by keeping those with the highest feature importance score from RF
  • filterByVar(): function to reduce number of features by keeping those with the largest variance

Modeling

• fit_models.R: functions for fitting various models while returning a consistent output that includes the 1) fitted predictions, 2) the test predictions, 3) the model fit, 4) a feature importance data frame, and 5) the estimated support

  • fitLM(): fit a linear model
  • fitLasso(): fit a Lasso model
  • fitRF(): fit a random forest
  • fitXGB(): fit an xgboost model

• eval_functions.R: functions for evaluating prediction models

  • evalPreds(): function to calculate prediction error between observed and predicted responses, as measured by MSE, $R^2$, MAE, correlation, (balanced) classification error, AUC, and/or PR
  • evalPermTest(): function to perform permutation test for prediction error between observed and predicted responses

Plots and Tables

• ggplot_themes.R: create customized ggplot theme

  • prettyGGplotTheme(): customized ggplot theme
  • prettyGGplotColor(): customized ggplot color theme
  • prettyGGplotFill(): customized ggplot fill theme

• ggplot_wrappers.R: convenient wrapper functions around common ggplot geometric objects (geoms) and adds custom ggplot theme

  • plotBarplot()
  • plotBoxplot()
  • plotDensity()
  • plotHistogram()
  • plotLine()
  • plotScatter()

• eda_plotting_functions.R: functions to generate common EDA plots

  • plotPairs()
  • plotPCA()
  • plotHeatmap()

• table_wrappers.R: convenient wrapper functions to make tables using kable and datatable

  • prettyKable(): function to make nice kable tables with options to bold certain entries
  • prettyDT(): function to make nice DT tables with options to bold certain entries

Rmarkdown Templates

• rmd_functions.R: Rmarkdown helper functions

  • subchunkify(): function to allow for multiple plots of different sizes and captions within single R code chunk in Rmd

• rmd_templates/: Rmarkdown templates for simulation results

  • custom_rmd_theme.css: custom css style
  • parameterized_results_plots.Rmd: generates html document with plots from specified folder (inputted as a parameter)
  • shiny_results_plots.Rmd: generates shiny interactive document with plots from specified folder

Miscellaneous

• numerical_checks.R: functions to conduct common numerical checks
  • isEqual(): function to check for equality between two matrices A and B up to given numerical tolerance
  • isOrthonormal(): function to check if matrix is orthonormal up to given numerical tolerance