R/myLinearRegression.R

In alexMoore190/myLinearPackage: Builds a multivariate linear model

#' myLinearRegression
#' 
#' 
#' This function builds a linear regression model and returns the coefficients 
#' and their p-values. However, it provides the option to only regress using a 
#' certain set of rows within a dataset. This makes it very useful for passing 
#' your test/train splits or crossvalidation splits into the model. It also 
#' returns a scatterplot matrix if you have fewer than five independent 
#' variables. Created by Alex Moore.
#' 
#' @param y The dependent variable as a vector
#' @param x The independent variables as a matrix
#' @param sub The index of rows/observations you'll use for building the model
#' @return A dataframe with the coefficients and p-values for the model 
#' @return A scatterplot matrix if you have less than 5 independent variables.
#' @export
#' @examples myLinearRegression(mtcars$mpg, mtcars[ , 2:4], 1:20)

myLinearRegression <- function(y, x, sub) {
  
  # Load the appropriate packages
  library(GGally)
  library(ggplot2)
  library(xtable)
  
  # Taking out observations so only the values from sub remain
  # Then binding x and y together for a single dataframe
  x <- x[sub, ]
  y <- y[sub]
  xy <- cbind(y, x)
  
  
  if(length(colnames(x)) < 5) {
    
    # Build the plot!
    print(ggpairs(xy, title = "Scatterplot Matrix") + 
            theme_bw())
    
  } else { # ends the first if statement
    
    print("Too many variables to plot")
  } # ends the else statement
  
  # Build the model
  modfit_xy <- lm(y ~ ., xy)
  
  # Get the summary table for the coefficients with the right information
  df_modfit_xy <- as.data.frame(summary(modfit_xy)$coefficients)
  
  # Make the table smaller so it only contains the columns of interest.
  df_modfit_xy <- df_modfit_xy[ , c(1, 4)]
  
  
  # I'm returning this as a dataframe instead of a list because a dataframe 
  # seems way more practical than a list and it contains all the information we 
  # need.
  return(df_modfit_xy)
  
} # Ends the function