From 12cd50bc3f3564b6b43c970147c2e563c6aa30a8 Mon Sep 17 00:00:00 2001
From: Erik-Jan van Kesteren <erikjanvankesteren@pm.me>
Date: Mon, 15 Apr 2024 15:39:19 +0200
Subject: [PATCH] Update 04_exercise.R

---
 04_exercise.R | 47 +++++++++++++++++++++++++++++++++++------------
 1 file changed, 35 insertions(+), 12 deletions(-)

diff --git a/04_exercise.R b/04_exercise.R
index 274b9c2..115fc09 100644
--- a/04_exercise.R
+++ b/04_exercise.R
@@ -1,21 +1,44 @@
 # Exercise time!
+
+# The assignment.
+# Create a plot with on the x-axis the three different contract types,
+# and on the y-axis the average wage per unit time (wage per hour). 
+# Add 95% confidence intervals (+- 1.96 * standard error of the mean)
 library(tidyverse)
 library(haven)
-spfns <- list.files("fake_cbs_data/Spolis/", full.names = TRUE)
+spolis_loc <- "fake_cbs_data/Spolis/SPOLISBUS2022V2.sav"
+
+# the plot for 100k rows below. With more samples
+# we can bring the s.e. down to see if there is a
+# significant difference between these items
+df_example <- read_spss(spolis_loc, n_max = 1e5) 
+
+df_example |>  
+  summarize(
+    mean   = mean(SBASISLOON / pmax(1, SBASISUREN)), 
+    stderr = sd(SBASISLOON / pmax(1, SBASISUREN)) / sqrt(n()),
+    lower  = mean - 1.96*stderr,
+    upper  = mean + 1.96*stderr,
+    .by = SCONTRACTSOORT
+  ) |> 
+  ggplot(aes(
+    x = as_factor(SCONTRACTSOORT, levels = "labels"),
+    y = mean,
+    ymax = upper,
+    ymin = lower
+  )) +
+  geom_pointrange() +
+  labs(
+    x = "Contract type",
+    y = "Average wage",
+    title = "Average wage per unit time for different contract types."
+  ) +
+  theme_linedraw()
 
 
-bigtab <- read_spss(spfns[1])
-for (fn in spfns[-1]) {
-  cat("reading", fn, "\r")
-  bigtab <- bind_rows(bigtab, read_spss(fn))
-}
+# Use your skills to do this for the whole data without loading it all
+# in at once! What is your conclusion?
 
-write_rds(bigtab, "bigtab.rds")
 
 
-N <- nrow(bigtab)
-bigtab$IKVID <- as.character(round(runif(N, min = 100000000000, max = 999999999999)))
-bigtab$SDATUMAANVANGIKO <- as.Date("2013-01-01")
-bigtab$SDATUMEINDEIKO <- as.Date("2022-08-23")
 
-write_sav(data = bigtab, path = "fake_cbs_data/Spolis/SPOLISBUS2022V2.sav")