Weekly schedule (R)

If not otherwise indicated, you can run each command below in R by creating these objects …

packages <- 
  c("tidyverse", "Rtsne", "igraph", "tidytext", "sf", "topicmodels",
    "knitr")

package <- sample(packages, 1)
  
numbers <- sample(1:1000, 100)

other_numbers <- sample(1:1000, 100)

factors <- factor(sample(c("red","blue"), 100, replace = T))

other_factors <- factor(sample(c("green","yellow"), 100, replace = T))

dataframe <- data.frame(numbers, other_numbers, factors, other_factors)
  
matrix <- cbind(numbers, other_numbers)

adjmatrix <- matrix(rbinom(100, 1, prob = .5), ncol = 10)

edglist <- matrix(sample(1:10, 100, replace = T), ncol = 2)

and installing/loading these packages first …

install.packages(packages)

lapply(packages, require, character.only = TRUE)

Week 01

• Launching RStudio

• Installing packages

  • install.packages(packages)

• Loading packages

  • library(package) or

  • require(package)

• Running simple calculations

  • 1 + 2 and

  • 2 - 3 and

  • 3 * 4 and

  • 4 / 5

• Creating objects

  • hello <- "hi" or hello = "hi"

• Using functions

  • function_name(arguments) [don’t run]

  • package::function_name(arguments) [don’t run]

• Loading data

  • load(file) [don’t run]

  • read.csv(file) [don’t run]

• Making sense of data (and data vector, matrices, data frames)

  • Identifying type of R object

    • class(dataframe)

  • Identifying length of vector

    • length(numbers)

  • Identifying number of rows (observations)

    • nrow(dataframe)

  • Identifying number of columns

    • ncol(matrix)

  • Identifying number of rows and columns

    • dim(matrix)

Week 02

• Getting the first part of matrix-like (rectangular) object

  • head(dataframe)

• Getting the column names of a matrix-like (rectangular) object

  • colnames(dataframe)

• Accessing variables inside dataframes

  • dataframe$numbers

• Computing means

  • mean(numbers)

Week 03

• Creating a ggplot with ggplot2

  • Using aesthetic mappings

    • ggplot2::ggplot(dataframe, aes(x = numbers, y = factors))

  • Choosing and using different geoms

    • Using geom_density

      • ggplot(dataframe, aes(numbers)) + ggplot2::geom_density()

    • Using geom_histogram

      • ggplot(dataframe, aes(numbers)) + geom_histogram()

    • Using geom_bar

      • ggplot(dataframe, aes(x = factors)) + geom_bar()

    • Using geom_point

      • ggplot(dataframe, aes(x = numbers, y = other_numbers)) + geom_point()

    • Using geom_smooth

      • ggplot(dataframe, aes(x = numbers, y = other_numbers)) + geom_smooth()

    • Using geom_boxplot

      • ggplot(dataframe, aes(x = factors, y = numbers)) + geom_boxplot()

    • Using geom_line

      • ggplot(dataframe, aes(x = numbers, y = other_numbers)) + geom_line()

  • Using scales

    • ggplot(dataframe, aes(x = numbers, y = other_numbers)) + scale_y_continuous()

  • Using facets

  ggplot(dataframe, aes(x = numbers, y = other_numbers)) + facet_grid()

Week 04

• Using pipes (%>%)

  • dataframe %>% dplyr::mutate(division = numbers / other_numbers)

• Manipulating dataframes with dplyr

  • Filtering observations

    • dplyr::filter(dataframe, numbers > 50, factors == "red")

  • Selecting variables

    • dplyr::select(dataframe, numbers, factors)

  • Creating new variables

    • dplyr::mutate(dataframe, division = numbers / other_numbers)

  • Grouping observation

    • dplyr::group_by(dataframe, factors)

  • Summarising variables

    • dplyr::summarise(dataframe, n = n())

  • Row-wise operations

    • dataframe %>% dplyr::rowwise() %>% dplyr::mutate(m = mean(c(numbers, other_numbers)))

Week 05

• Using relational operators

  • 3 == 3 and

  • TRUE != FALSE and

  • 3 < 4 and

  • 4 > 3

• Creating variables with ifelse

  • ifelse(factors == "red", 1, 0)

• Subsetting variables

  • numbers[numbers > 50]

• Subsetting data frames

  • dataframe[1,] or

  • dataframe[,1] or

• Reporting with R Markdown

  • knitr::kable(dataframe, caption = "My caption")

  • knitr::kable(dataframe, caption = "My caption")

Week 06

• Performing principal component analysis

  • prcomp(matrix, scale = TRUE)

• Performing t-SNE analysis

  • Rtsne::Rtsne(matrix, perplexity = 10)

Week 07

• Creating frequency tables

  • table(factors)

• Creating tables of proportions

  • prop.table(table(factors))

• Understanding missing values

  • is.na(factors) and

  • is.null(numbers)

• Handling missing values

  • table(factors, exclude = NULL) and

  • na.omit(factors)

• Creating two-way frequency tables

  • table(factors, other_factors)

• Creating two-way tables of proportions

  • prop.table(table(factors, other_factors))

• Computing descriptive statistics

  • Mean

    • mean(numbers)

  • Median

    • median(numbers)

  • Standard deviation

    • sd(numbers)

  • Variance

    • var(numbers)

• Computing correlations

  • cor(numbers, other_numbers)

Week 08

• Taking a random sample

  • sample(numbers, size = 10)

• Taking a random sample from a normal distribution

  • rnorm(10)

• Computing probability of normal random variables

  • pnorm(-1.96)

• Using for loops

  • for(i in 1:10) {print(i)}

• Computing absolute value

  • abs(-10)

Week 09

• Computing 95% confidence interval for sample means

  • mean(numbers) - 1.96 * sqrt(var(numbers) / length(numbers)) and

  • mean(numbers) + 1.96 * sqrt(var(numbers) / length(numbers))

• Computing 95% confidence interval for difference-in-means estimators

  • mean(numbers) - mean(other_numbers) - 1.96 * sqrt(var(numbers) / length(numbers) + var(other_numbers) / length(other_numbers)) and

  • mean(numbers) - mean(other_numbers) + 1.96 * sqrt(var(numbers) / length(numbers) + var(other_numbers) / length(other_numbers))

• Hypothesis testing with the difference-in-means estimators

  • z_obs <- (mean(numbers) - mean(other_numbers)) / sqrt(var(numbers) / length(numbers) + var(other_numbers) / length(other_numbers)) and

  • 2 * pnorm(-abs(z_obs))

Week 10

• Fitting simple linear regression models

  • fit <- lm(numbers ~ other_numbers, data = dataframe)

• Using natural logarithmic transformations

    * `log(numbers)` or
  
    * `lm(log(numbers) ~ other_numbers, data = dataframe)` or
  

• Hypothesis testing with estimated regression coefficients

  • summary(fit)$coef

• Reporting linear model results

  • summary(fit)

Week 11

• Fitting multiple linear regression models

  • lm(numbers ~ other_numbers + factors + other_factors)

• Computing confidence interval for predicted outcomes

  • predict(fit, newdata = data.frame(other_numbers = 100), interval = "confidence")

Week 12

• Fitting logistic regression models

  • glm(factors ~ numbers + other_numbers, family = 'binomial')

Week 12 & 13 (advanced topics)

• Analysing network data

  • Creating networks

    • g <- igraph::graph_from_adjacency_matrix(adjmatrix)

    • g <- igraph::graph_from_edgelist(edglist)

  • Visualising networks

    • plot(g)

  • Computing network statistics

    • Size (global)

      • igraph::vcount(g) and

      • igraph::ecount(g)

    • Degree (local)

      • igraph::degree(g)

    • Closeness (local)

      • igraph::closeness(g)

    • Betweenness (local)

      • igraph::betweenness(g)

  • Computing network communities

    • igraph::cluster_optimal(g)

• Analysing spatial data

  • Loading spatial data

    • data.sf <- sf::read_sf(file) [don’t run]

  • Visualising spatial data

    • leaflet::leaflet(sf) + leaflet::addTiles() [don’t run] or

    • ggplot2::ggplot(sf) + geom_sf() [don’t run]

  • Manipulating spatial data

    • sf::st_buffer(sf, dist = 150) [don’t run]

    • sf::st_intersects(sf) [don’t run]

  • Getting spatial features from OpenStreetMap

    • query <- osmdata::opq(bbox = c(13.30,45.89,13.31,45.91))

    • osm_feature <- osmdata::add_osm_feature(query, key = 'building')

    • sf <- osmdata::osmdata_sf(osm_feature)

• Analysing text data

  • Tidying text data

  • Analysing word frequencies

  text_df %>% unnest_tokens(word, text) [don’t run]

  tidy_df %>% anti_join(stop_words) [don’t run]

  • Analysing sentiment of texts

  tidy_books %>% dplyr::inner_join(tidytext::get_sentiments("bing")) [don’t run]

  • Analysing relationships between words

  • Identifying topics in texts

  topicmodels::LDA(dtm, k = 10) [don’t run]