In hmumme/bmi585hmumme: What the Package Does (One Line, Title Case)

This Markdown file contains example uses for all the functions in the package bmi585hmumme. You can download the package with the following command in the R Console. Uncomment the following if you have not installed the package to your system.

#devtools::install_github("hmumme/bmi585hmumme")

Load the package and set up the Markdown file:

knitr::opts_chunk$set(echo = TRUE)
library(bmi585hmumme)

The following functions are broken down by section; each section represents the assignment the function came from. These examples can also be found in the help documentation for each function; type ?functionName to retrieve this.

newFunctions.R

unscale(x)

This function takes a numerical object from scale(), reverses the scaling and centering done (if any), and returns the un-scaled numerical object.

xS = scale(matrix(1:10, ncol = 2)) # scale a matrix
unscale(x = xS) # un-scale to get the original matrix

pcApprox(x,npc)

This function approximates x feature data using npc number of principle components.

x = matrix(c(4,5,2,10,2,8,4,5,6), ncol = 3) # feature data matrix
pcApprox(x, npc = 1) # approximate using 1 principle component

pcLollipop(x)

This function creates lollipop plots of all the principal component loadings in feature data x (must be numerical data frame or tibble). Each principle component will have its own plot of values for each variable in the data x.

x = data.frame(data = matrix(c(4,5,2,10,2,8,4,5,6), ncol = 3)) # feature data frame
pcLollipop(x) # create lollipop plots

quiz02Functions.R

sensitivity(pred,truth)

This function calculates the sensitivity from predicted and true vectors. These vectors can either be 0s (negatives) and 1s (positives) or boolean.

sensitivity(c(0,1,0),c(1,1,1)) # calculates sensitivity

specificity(pred,truth)

This function calculates the specificity from predicted and true vectors. These vectors can either be 0s (negatives) and 1s (positives) or boolean.

specificity(c(0,1,1),c(0,0,1)) # calculates specificity

accuracy(pred,truth)

This function calculates the accuracy from predicted and true vectors. These vectors can either be 0s (negatives) and 1s (positives) or boolean.

accuracy(c(0,1,1),c(1,1,1)) # calculates accuracy

ppv(pred,truth)

This function calculates the PPV from predicted and true vectors. These vectors can either be 0s (negatives) and 1s (positives) or boolean.

ppv(c(1,1,0),c(0,1,1)) # calculates PPV

f1(pred,truth)

This function calculates the F1 score from predicted and true vectors. These vectors can either be 0s (negatives) and 1s (positives) or boolean.

f1(c(1,1,1),c(1,1,0)) # calculates f1 score

hw04Functions.R

boxMuller(n)

This function use the Box-Muller transformation to generate 2 vectors of random, normally distributed n values.

boxMuller(5) # generates two random, normally distributed vectors with 5 values each

hw06Functions.R

twoSidedT(t,n)

This function calculates the p-value from a two-sided t-test statistic and the number of samples n.

twoSidedT(3, 5) # calculates p-value

twoSizedZ(z)

This function calculates the p-value from a two-sided z score.

twoSidedZ(3) # calculates p-value

effectSize(x,g)

This function calculates the effect size from vector of values x and grouping variable g.

effectSize(c(1.42, 8.10, 9.22, 20.92, 5.67, 0.45), c(0,0,1,1,0)) # calculates effect size

welchT(x,y)

This function performs a Welch's T test to compare the means of vector x and vector y. It will return a vector of length 5 with Welch's T statistic, degrees of freedom, p-value, mean of group x, and mean of group y.

welchT(c(1.42, 8.10, 9.22, 20.92, 5.67, 0.45), c(1.99, 8.09, 9.13, 9.10, 5.55, 0.11)) # calculate welch t-test statistics

hw07Functions.R

minimumN(d)

This function finds the minimum number of samples needed for effect size d and power = 0.8, returns number of samples rounded up to nearest integer.

minimumN(1.3) # calculate number of samples needed

chiSquareCounts(tib,x,y)

This function performs a chi square test of homogeneity between data in columns x and y of tib. Returns a p-value.

data = dplyr::tibble("sex" = c(1,0,1),"group" = c("A","B","C"), "age" = c(40,15,33), "height" = c(63, 70, 68)) # create data
chiSquareCounts(data, "age", "height") # perform test

postHocPower(d,n1,n2)

This function estimates post hoc power from effect size and number of samples using 1000 simulations.

postHocPower(d = 0.5, n1 = 20, n2 = 22) # estimate power

hw08Functions.R

bhAdjust(p)

This function tests for significant p-values using Bonferroni-Holm adjustment. Returns logical vector showing which p-values are statistically significant after the BH adjustment.

bhAdjust(c(0.05, 0.0005, 0.01, 0.0225, 0.025)) # find which p-values are significant after BH adjustment

fdrAdjust(p)

This function tests for significant p-values using FDR adjustment. Returns logical vector showing which p-values are statistically significant after the FDR adjustment.

fdrAdjust(c(0.05, 0.0005, 0.01, 0.0225, 0.025)) # find which p-values are significant after fdr adjustment

hw11Functions.R

r2(pred,truth)

This function calculates the R-Squared value from two numerical vectors of predicted/true values.

yTrue = c(0,1,1,1) # true values
r2(pred = c(0,1,0.7,1), truth = yTrue) # find R^2

adjR2(pred,truth)

This function calculates the adjusted R-Squared value from two numerical vectors of predicted/true values and the number of predictors used d.

yTrue = c(0,1,1,1) # true values
adjR2(pred = c(0,1,0.7,1), truth = yTrue, d = 2) # find adjusted R^2

hmumme/bmi585hmumme documentation built on Dec. 20, 2021, 4:46 p.m.