R Functions and Procedures Summarized
--------------------------------------------------------------------------------
Important Note: R is case sensitive!
For an example of any of the following functions:
> example(function_name)
...and if R says "Waiting to confirm page change..." press Enter.
--------------------------------------------------------------------------------
simple arithmetic performed at the command line (a, b may be numbers or vectors)
     a + b means add a and b
     a - b means subtract a and b
     a * b means multiply a and b
     a / b means divide a by b
     (a + b)/b means grouping works as you might expect; this equals a/b + 1
--------------------------------------------------------------------------------
abline(a=intercept, b=slope) - draw a line on an existing graph
abline(h=value) - draw a horizontal line on an existing graph
abline(v=value) - draw a vertical line on an existing graph
abline(lm_results) - plot the regression line on an existing scatterplot
addmargins(contingency_table_name) - print contingency table with marginal sums
anova(lm_results) - see the anova on stored results of a linear regression
anova(lm_results1, lm_results2) - test two stored lm models against each other
aov(formula) - ANOVA
attach(dataframe_name) - CAREFULLY(!!!) attach a dataframe to search path
barplot(table_name) - plot a bar graph of values in table (must be tabled first)
barplot(2D_table_name, beside=T, legend=T) - side-by-side bar graph
bartlett.test() - test for homogeneity of variance
binom.test(number_of_successes, N, p=null prob.) - exact binomial test
boxplot(variable_name) - plot a boxplot of a numerical variable
boxplot(DV ~ IV) - plot boxplots at each level of an independent variable
c(data_values) - concatenate; group values into a variable or vector
chisq.test(vector_name, p=vector_null_probs.) - chi square goodness of fit test
chisq.test(table_name) - chi square significance test of indpendence
chisq.test_result$expected - examine expected frequencies
chisq.test_result$residuals - examine residuals
class(object_name) - find out what kind of object R considers this to be
colnames(tablelike_object) - create or get column names; also names()
confidence intervals - to get a CI, do the corresponding significance test
cor(variable_name1, variable_name2) - Pearson's r between numerical variables
cor(var_name1, var_name2, use="pairwise.complete") - if there are missing values
cor(var_name1, var_name2, method="spearman") - Spearman correlation
cor(dataframe_name) - get a correlation matrix of all numerical variables
cor.test(var_name1, var_name2) - significance test for r sign. diff. from zero
cor.test(var_name1, var_name2, alternative="more" or "less") - one-sided tests
curve(function_of_x, from=value, to=value) - plot an equation
data.frame(vector1, vector2, vector3, ...) - create a dataframe from vectors
dataframe_name[i,] - get row i of the dataframe
dataframe_name[i:j,] - get rows i through j of the dataframe
dataframe_name[,k] - get column k of the dataframe; or add column to dataframe
dataframe_name$variable_name - get column of dataframe with given variable name
dbinom() - binomial probability function
detach(dataframe_name) - remove dataframe from search path
dim(object_name) - get the dimensions of a table, matrix, or dataframe
dir() - view files in the working directory
exp(value) - raise e to the power of value; natural antilog
fisher.test(contingency_table_name) - Fisher's exact test
fivenum(variable_name) - five number summary of a numerical variable
function() - program your own functions into R; see sem
getwd() - find out what your working directory is
hist(variable_name) - plot a histogram of a numerical variable
hist(var_name, breaks=vector_of_desired breaks) - with specified break points
interaction.plot() - draw a graph of an interaction between two variables
IQR(variable_name) - get the interquartile range using quantile() type=7
kruskal.test() - nonparametric alternative to oneway ANOVA
legend() - add a legend to an existing plot; complicated
length(variable_name) - get the no. of cases in a variable (n); chokes on NAs
levels(variable_name) - get the levels of a categorical variable
library(library_name) - load an optional library, for example MASS
lines(specification_of_line) - draw a line on an existing graph
lm(DV ~ IV) - linear regression of DV on IV using least squares
lm(DV ~ IV + I(IV^2) + I(IV^3) + ...) - polynomial regression
lm(DV ~ IV1 + IV2 + IV3 + ...) - multiple linear regression
lm_model$residuals - examine the residuals from a stored lm model
load("location_of_previously_saved_file") - load a "saved" file; see save()
log(value) - take the natural log of a number; log(value, base)=log to any base
log(variable_name) - do a log transform on a numerical variable (base-e)
log10(variable_name) - a base-10 log transform
ls() - list the objects in the workspace; also objects(); character(0)=nothing
mad(variable_name) - get the median absolute deviation from the median
margin.table(contingency_table_name, 1) - get the row marginal sums
margin.table(contingency_table_name, 2) - get the column marginal sums
matrix(vector_name, nrow=value, ncol=value) - create a matrix from a vector
max(variable_name) - get maximum value in a numerical variable
mean(variable_name) - get mean of a numerical variable; chokes on NAs
mean(variable_name, na.rm=T) - mean of variable with missing values
median(variable_name) - get median of the numerical variable; na.rm is available
min(variable_name) - get the minimum value in a numerical variable
NA - a missing value (R's missing value code)
names() - can be used to name variables or to retrieve the names of variables
oneway.test(formula) - one-way betw. subjects ANOVA
pbinom() - binomial cumulative probability function
pie(table_name) - pot a piechart of values in table (must be tabled first)
plot(predictor_var, response_var) - scatterplot of two numerical variables
plot(var1, var2, log="y") - plot with a logarithmic y-axis; fails if any y<1
plot(dataframe_name) - plot a matrix of scatterplots of numerical variables
plot(density(variable_name)) - kernel density smoother for numerical variable
plot(lm_results) - plot four regression diagnostic plots from stored lm results
plot.new() - open the graphics device for plotting
pnorm(z) - get the area under the unit normal curve that is below z
points(vector_of_x-values, vector_of_y-values) - plot points on a graph
prop.table(table_name) - convert tabled frequencies to proportions of N
prop.table(table_name, 1) - convert tabled freqs. to proportions by rows
prop.table(table_name, 2) - convert tabled freqs. to proportions by columns
prop.test(number_of_hits, N, p=null_prob.) - one-proportion test
prop.test(vector_of_hits, vector_of_Ns) - multiple-proportion test
qqline(variable_name) - draw the normal line on an existing qqnorm() plot
qqnorm(variable_name) - plot a normal probability plot of a numerical variable
qqplot(variable1, variable2) - compare distributions of two variables
quantile(variable_name, probs=vector_of_probs.) - get desired quantiles
quantile(variable_name, probs=c(1/4, 3/4), type=2) - get textbook Q1 and Q3
range(variable_name) - reports min() and max()
rbind(vector1, vector2, vector3, ...) - bind vectors into a table by rows
read.csv("file_location") - read in a csv formated data file
read.table("file_location", header=T) - read in tabled data with headers
rm(object_name) - remove an object from the workspace; also remove()
rbinom() - generates random binomially distributed values
rep(data_vector, freq_vector) - creates a vector with repeating elements
rnorm() - generates random normally distributed values
rownames(tablelike_object) - create or get row names; see colnames()
save(R_object, file="filename") - save a data object to a file in working dir.
scan() - accept data input directly from the keyboard
scatter.smooth(predictor_var, response_var) - scatterplot with lowess smoother
sd(variable_name) - get the sample standard deviation of a numerical variable
search() - as R where it is looking for stuff (the search path)
sem=function(x) {sd(x)/sqrt(length(x))} - quick and dirty sem function
seq(from=value, to=value, by=value) - create a regular sequence of numbers
setwd() - set or change the working dircetory (better to use the menus)
sort(variable_name) - sort values in ascending order
sort(variable_name, decreasing=T) - sort values in descending order
source("location_of_script") - read in and execute a text file of R commands
sqrt(number or vector_name) - take the square root of number or numbers
stem(variable_name) - get a stem and leaf display of a numerical variable
step() - automated forward, backward, or stepwise regression; not a good idea!
str(object_name) - find out the structure of an R object
subset() - subset a dataframe
sum(vector_name) - sum the values in a vector; na.rm is available
summary(object_name) - summarize an R object; summary stats from a variable
summary(lm_results) - get a detailed summary of stored linear regression results
table(variable_name) - create a 1D frequency table of a variable
table(var_name1, var_name2) - create a contingency table with var_name1 in rows
tapply(DV, IV, function) - calculation a function of a DV at each level of an IV
t.test(variable_name, mu=null_mean) - single-sample t test
t.test(var_name, mu=null_mean, alternative="more" or "less") - one-sided test
t.test(variable1, variable2) - two-sample t test with Welch correction
t.test(var1, var2, var.equal=T) - no Welch correction (pooled variance t test)
t.test(var1, var2, alternative="more" or "less") - one-sided test (gp1 - gp2)
t.test(DV ~ IV) - two-sample t test using formula interface (from a dataframe)
t.test(... , paired=T) - related samples t test
text(x-coord, y-coord, "string of text") - add text to an existing plot
title(main="text string", xlab="text string", ylab="text string") - add titles
     and axis labels to an existing plot
update(lm_model, ~.-variable_name) - update an lm model by deleting a variable
var(variable_name) - get the sample variance of a numerical variable
var.test(vector1, vector2) - test for a significant difference betw. variances
vector_name[i] - get the value at position i of the vector
vector_name[i:j] - get values at positions i through j of the vector
which(variable_name .test. value) - find out which values pass the test; .test.
     may be < (less than), > (greater than), == (equals), != (not equal)
wilcox.test(DV ~ IV) - commonly called the Mann-Whitney U test
wilcox.test(DV ~ IV, paired=T) - Wilcoxin test for paired scores
with(dataframe-name, function) - specify dataframe to be used with a function
xtabs(formula, data=dataframe_name) - crosstabulation; see handout for formula
z.test - doesn't come with the default packages; can be added in or programmed
--------------------------------------------------------------------------------