In MGallow/statr: Matt Galloway Personal R Package

knitr::opts_chunk$set(echo = TRUE)

Overview

statr is a personal R package that I have created for organizational/convenience purposes. This project is purely experimental! A (possibly incomplete) list of functions contained in the package can be found below:

tidy() tidy's R package code and updates documentation
CVsplit() splits data objects into training and testing sets
data_gen() generates data for linear regression settings
dense() generates dense matrices
denseQR() generates dense matrices via spectral decomposition
tridiag() generates tri-diagonal matrices
derivative() approximates the derivative for a given function
diagnostic() creates diagnostic plots using ggplot (residual and QQ)
dsearch() is a dichotomous search algorithm for minimizing a univariate function
bsearch() is a bi-section search algorithm for minimizing a univariate function
LASSO() calculates lasso regression coefficient with optimal tuning
RIDGE() calculates ridge regression coefficient with optimal tuning
scatter() creates a scatterplot using ggplot

See vignette or manual.

Installation

The easiest way to install is from the development version from Github:

# install.packages("devtools")
devtools::install_github("MGallow/statr")

If there are any issues/bugs, please let me know: github. You can also contact me via my website. Contributions are welcome!

Usage

library(statr)
library(magrittr)

# we will use the iris data set
X = dplyr::select(iris, -c(Species, Sepal.Length)) %>% as.matrix
y = dplyr::select(iris, Sepal.Length) %>% as.matrix
y_class = ifelse(dplyr::select(iris, Species) == "setosa", 1, 0)

# let us split the data for testing and training
CV = CVsplit(X, y)

# we can do some exploratory analysis
# plot Sepal.Length v Sepal.Width
iris %>% scatter(Sepal.Length, Sepal.Width)

# plot diagnostic plots
iris %>% diagnostic(Sepal.Length, Sepal.Width)

# use the training data to fit ridge regression
RIDGE(CV$X.train, CV$Y.train)

# or lasso regression
statr::LASSO(CV$X.train, CV$Y.train)

# we can also generate our own data
data = data_gen(p = 10, r = 5, n = 100)
CV = CVsplit(data$X, data$Y)

# and again fit a ridge regression
statr::RIDGE(CV$X.train, CV$Y.train)

# we can also generate random matrices with may be useful
# for other applications
# tridiagonal matrices
tridiag(p = 5)$Omega %>% round(5)

# dense matrices
dense(p = 5)$Omega %>% round(5)

# compound symmetric matrices
compound(p = 5)$Omega %>% round(5)