Cheatsheet for Data in the Wild Tutorials

Basic R Commands

  • Basic arithmetic

    • Examples: 5*6, sqrt(9), abs(-3)

    • Type help(command) to find information about any command

  • Create variables: Use <- or = to assign values to a variable.

    • Example: x <- 7

  • Create a vector:

    vector <- c(1,3,7)
vector
    [1] 1 3 7

  • Create a data frame: data.frame(____)

    df <- data.frame(var_1 = c(1,2,3) , var_2 = c("cat", "dog", "fish"))
df
      var_1 var_2
1     1   cat
2     2   dog
3     3  fish

Basic Statistics

  • mean(vector) : Calculates the mean of a given set of values.

  • median(vector) : Calculates the median of a given set of values.

  • sd(vector) : Calculates the standard deviation of a given set of values.

  • nrow(data): Calculates the total number of rows in a dataset

  • na.rm = TRUE : Remove NA values. Add this as an argument to any of the statistics calculations. E.g. mean(vector, na.rm=TRUE)

Working with a data set in the tidyverse

We use the tidyverse to analyze data in this cheatsheet. To install run the following code:

install.packages("tidyverse")

After installing, to load the tidyverse run:

library(tidyverse)

Uploading and viewing a data set

  • dataFrame <- read_csv("myCSV.csv") : Creates a data frame from a file called myCSV.csv

  • write_csv(dataFrame, "myCSV.csv"): Creates a csv from a data frame

  • view(dataFrame): Enter view-mode to see the entire data frame

  • str(dataFrame): Gives the structure of data frame

  • glimpse(dataFrame): Take a quick look at a data frame

  • summary(dataFrame): Returns min, max, mean, meadian, 1st/3rd quartiles for all vectors in dataFrame

  • dataFrame$columnName: Calls up specific column from a data frame

  • unique(vector): Find the number of unique values in a column/vector

  • length(): Counts the number of entries in a column/vector

Rearranging & Summarizing data

  • Use logical operators to combine conditions: & (and), | (or), ! (not)

  • Use comparison operators to describe conditions: <, >, ==, !=, <=, >=

  • group_by(.data, column): Takes a dataset and groups it by a column/variable

  • summarize(.data, summaryStat = statistic formula): Takes a dataset and outputs summary statistics that you define.

    • n(): Calculates current group size. Can be used in summarize and group_by

    • Combine group_by and summarize using the pipe (|>) to see summary statistics for specific groups/variables.

    • The pipe: Use |> OR %>% to string functions and data together. Read as “AND THEN”.

    Example:

      dataFrame |> 
    group_by(firstColumn)|> 
    summarize(mean_of_secondColumn = mean(secondColumn), 
              standard_deviation_of_secondColumn = sd(secondColumn),
              number_in_each_group = n(), 
              percent = n()/nrow(dataFrame)*100)

  • filter(.data=myDataFrame, column=="some value"): Subsets dataframe, lets us pick rows of data based on “some value”, including a specific value, mean, median, etc.

  • select(.data, column(s)): Choose columns from a dataframe.

    • column1 , column2 : choose column1 and column2
    • column1:column5 : choose all columns between column1 and column5
    • -column1: choose all columns except column1

  • mutate(data, newColumn = yourFormula): Creates a new column in a dataset defined by a calculation you input.

  • arrange(data, column): Rearrange data into numerical order by a specific column

Plotting with GGPlot

Resources:

GGPlot Basics:

  • Basic structure:

    ggplot(data, mapping=aes()) +
      geom_function()

  • Connect different properties using a +

  • Basic Components:

    • Data: data=dataSet: Define your data set

    • Aesthetics: mapping = aes(variables): Define the variables. Can also specify color/fill for your graph and geometries.

    • Geometry: geom_GeometricObject(): Define the type of plot

Geometries

  • geom_histogram(): Creates a histogram

    • Syntax: geom_histogram(bins=X) Specify number of bins

  • geom_point(): Creates points (scatterplot) for each data point

    • Can specify color = ___ and shape = ____

  • geom_bar() : Creates a bar graph

    • Syntax:
      • geom_bar(): Aggregates data for you
      • geom_bar(stat = "identity"): Creates a bar graph with pre-aggregated data that you input
      • geom_bar(position="___") Can choose “stack” (bars on top of each other), “dodge” (bars side by side) or “fill” (bars stacked, scaled to 100%).
      • Add error bars: geom_errorbar(mapping=aes(ymin, ymax), width)

  • geom_boxplot(): Creates a boxplot

    • Add error bars: stat_boxplot(geom="errorbar")

  • geom_density(): Creates a density graph

    • Syntax: geom_density(adjust=X) X is a ratio, how smooth the graph is

  • geom_smooth(method="lm", se=FALSE): Creates a line of best fit

  • labs() : Add a title and axes labels to your graph

  • facet_wrap(~ variable): Create separate plots for each aspect of a given variable

  • And more!

Other useful GGPlot functions:

  • install.packages("ggplot2") and library(ggplot2): Install and load GGplot package. Note: ggplot2 is included in the tidyverse so you do NOT need to load both packages.

  • ggsave("yourTitleHere.jpg", currentPlotName): Saves “currentPlotName” as a jpg file called “yourTitleHere.jpg”.

Examples: Using GGPlot with other tidyverse functions

Use the pipe to string together functions and create tidy data before plotting.

The following examples use data from the palmerpenguins package which has been preloaded into this document. Click the green play button to see them run. Feel free to edit the code as well if you want to try out other functions!

Here’s a preview of the penguin data, and here is a link to more information about the data.

Basic bar graph of median flipper length on each island:

Boxplot of Adelie penguin flipper length, by sex:

Inferential Statistics

Basic statistical tests

  • t.test(data, dependentVariable ~ independentVariable): 2 sample t-test - compare 2 groups on a numerical measure

    • Requires: 2 samples of quantitative data

  • aov(data, dependentVariable ~ independentVariable): ANOVA (analysis of variance) test - compare means among groups

    • Requires: Independent variable - categorical; dependent variable - numerical

  • TukeyHSD(data): Tukey’s HSD - Post Hoc test which tells you which specific means differ from each other

Linear regression

  • cor(data$independentVariable, data$dependentVariable): Correlation coefficient - outputs a number between -1 and 1 which tells strength and direction of the relationship between 2 numeric variables

    • Requires 2 numeric variables

  • lm(data, dependentVariable ~ independentVariable): Fit a linear model

    • summary(yourLinearModel): View more details about a linear model including regression coefficients, p value for linear model coefficients, quartiles, residual standard error, F-statistic.