PCA on Titanic Dataset

library(ggplot2)
library(titanic)  


data("titanic_train")
titanic_data <- titanic_train

head(titanic_data)

## Calculates principal components with given formula on the data.
calc_pc <- function(formula, data) {
  print('Starting PCA')
  prcomp(formula, data)
}

predict_pc <- function(prcompValues, newdata, ...) {
  print('Predicting PC values')
  predict(prcompValues, newdata, ...)
}

show_pc <- function(data, formula) {
  print('Starting show PC')
  pca <- calc_pc(formula, data)

  print(pca)
  print(pca_plot(pca, 'PC1', 'PC2'))
  print(pca_plot(pca, 'PC3', 'PC4'))
  pca
}

pca_plot <- function(pca, byX = 'PC1', byY = 'PC2') {
  loadings <- data.frame(pca$rotation, .names = row.names(pca$rotation))

  theta <- seq(0, 2*pi, length.out = 100)
  circle <- data.frame(x = cos(theta), y = sin(theta))

  columnNameWithOffsetX <- paste(byX, '+ 0.2')
  columnNameWithOffsetY <- paste(byY, '- 0.1')
  p <- ggplot(circle, aes(x, y)) +
      geom_path() +
      geom_text(data=loadings,
          mapping=aes_string(x = columnNameWithOffsetX, y = columnNameWithOffsetY,
              label = '.names', color = '.names')) +
      geom_jitter(data = loadings, mapping = aes_string(x = byX, y = byY, color = '.names'), shape = 2) +
      coord_fixed(ratio = 1) +
      labs(x = byX, y = byY)
}


formula <- ~ Age + Fare + Pclass + SibSp + Parch

Call show_pc with the specified formula

show_pc(titanic_data, formula)

[1] "Starting show PC"
[1] "Starting PCA"
Standard deviations (1, .., p=5):
[1] 52.9413015 14.4605550  0.9747181  0.6956846  0.6359113

Rotation (n x k) = (5 x 5):
                PC1         PC2          PC3           PC4          PC5
Age     0.028489335 -0.99912896  0.025843692 -0.0004816311 -0.016172288
Fare    0.999546999  0.02854675 -0.003698066 -0.0013684373 -0.008679397
Pclass -0.008814850  0.01819365  0.107142164 -0.2551973250 -0.960721693
SibSp   0.002385886  0.02093118  0.766326815  0.6366812575 -0.083284949
Parch   0.003280366  0.01254088  0.632915632 -0.7276738260  0.264084401
Standard deviations (1, .., p=5):
[1] 52.9413015 14.4605550  0.9747181  0.6956846  0.6359113

Rotation (n x k) = (5 x 5):
                PC1         PC2          PC3           PC4          PC5
Age     0.028489335 -0.99912896  0.025843692 -0.0004816311 -0.016172288
Fare    0.999546999  0.02854675 -0.003698066 -0.0013684373 -0.008679397
Pclass -0.008814850  0.01819365  0.107142164 -0.2551973250 -0.960721693
SibSp   0.002385886  0.02093118  0.766326815  0.6366812575 -0.083284949
Parch   0.003280366  0.01254088  0.632915632 -0.7276738260  0.264084401