finished session3

Presentations-GeorgiaRS/202309/exercises-session2.R

@@ -1,3 +1,7 @@
+# Data
+small_business_2019     <- read.csv("data/small_business_2019.csv")
+small_business_2019_age <- read.csv("data/small_business_2019_age.csv")
+
 # Exercise 3
 library(dplyr)
 

Presentations-GeorgiaRS/202309/exercises-session3.R

@@ -0,0 +1,40 @@
+# Data
+small_business_2019_all <- read.csv("data/small_business_2019_all.csv")
+View(small_business_2019_all)
+
+# Exercise 1
+#install.packages("modelsummary")
+#install.packages("huxtable")
+
+# Exercise 2
+df_tbilisi_50 <- filter(small_business_2019,
+                        region == "Tbilisi") %>%
+  arrange(-income) %>%
+  filter(row_number() <= 50)
+
+# Exercise 3
+library(modelsummary)
+datasummary_skim(small_business_2019_all)
+
+# Exercise 4
+datasummary(
+  age + income + vat_liability ~ N + Mean + SD + Min + Max,
+  small_business_2019_all
+)
+
+# Exercise 5
+library(huxtable)
+stats_table <- datasummary_skim(small_business_2019_all, output = "huxtable")
+quick_xlsx(stats_table, file = "quick_stats.xlsx")
+
+# Exercise 6
+stats_table_custom <- stats_table %>%
+  set_header_rows(1, TRUE) %>%
+  set_header_cols(1, TRUE)  %>%
+  set_number_format(everywhere, 2:ncol(.), "%9.0f") %>%
+  set_align(1, everywhere, "center") %>%
+  theme_basic()
+quick_xlsx(
+  stats_table_custom,
+  file = "stats-custom.xlsx"
+)

Presentations-GeorgiaRS/202309/exercises-session4.R

@@ -0,0 +1,125 @@
+# Data
+small_business_2019_all <- read.csv("data/small_business_2019_all.csv")
+
+# Exercise 1a
+library(ggplot2)
+ggplot(small_business_2019_all) +
+  aes(x = taxperiod,
+      y = vat_liability) +
+  geom_col() +
+  labs(title = "Total VAT liability of small businesses in 2019 by month")
+
+# Exercise 1b
+ggplot(small_business_2019_all) +
+  aes(x = taxperiod,
+      y = vat_liability) +
+  geom_col() +
+  labs(title = "Total VAT liability of small businesses in 2019 by month",
+       # x-axis title
+       x = "Month",
+       # y-axis title
+       y = "Georgian Lari") +
+  # telling R not to break the x-axis
+  scale_x_continuous(breaks = 201901:201912) +
+  # centering plot title
+  theme(plot.title = element_text(hjust = 0.5))
+
+# Exercise 1c
+ggsave("vat_liability_small_2019.png",
+       width = 20,
+       height = 10,
+       units = "cm")
+
+# Exercise 2a
+df_group_month <- small_business_2019_all %>%
+  select(group, taxperiod, vat_liability) %>%
+  group_by(group, taxperiod) %>%
+  summarize(total = sum(vat_liability))
+
+# Exercise 2b
+ggplot(df_group_month) +
+  aes(x = taxperiod,
+      y = total) +
+  geom_line(aes(color = group)) +
+  labs(title = "Total VAT liability of small businesses in 2019 by experiment group",
+       x = "Month",
+       y = "Georgian Lari") +
+  scale_x_continuous(breaks = 201901:201912) +
+  theme(plot.title = element_text(hjust = 0.5))
+
+# Exercise 2c
+  ggplot(df_group_month) +
+    aes(x = taxperiod,
+        y = total) +
+    geom_line(aes(color = group)) +
+    labs(title = "Total VAT liability of small businesses in 2019 by experiment group",
+         x = "Month",
+         y = "Georgian Lari") +
+    scale_x_continuous(breaks = 201901:201912) +
+    theme(legend.text = element_text(size = 7),
+          axis.text.x = element_text(size = 6))
+
+# Exercise 2d
+  ggsave("vat_liability_small_2019_by_group.png",
+         width = 20,
+         height = 10,
+         units = "cm")
+
+# Exercise 3a
+df_month <- small_business_2019_all %>%
+  select(taxperiod, vat_liability) %>%
+  group_by(taxperiod) %>%
+  summarize(total = sum(vat_liability))
+
+# Exercise 3b
+ggplot(df_month) +
+  aes(x = taxperiod,
+      y = total) +
+  geom_col() +
+  geom_text(aes(label = total),
+            position = position_dodge(width = 1),
+            vjust = -0.5,
+            size = 3) +
+  labs(title = "Total VAT liability of small businesses in 2019 by month",
+       x = "Month",
+       y = "Georgian Lari") +
+  scale_x_continuous(breaks = 201901:201912) +
+  theme(plot.title = element_text(hjust = 0.5))
+
+# Exercise 3c
+  ggplot(df_month) +
+    aes(x = taxperiod,
+        y = total) +
+    geom_col() +
+    geom_text(aes(label = round(total)),
+              position = position_dodge(width = 1),
+              vjust = -0.5,
+              size = 3) +
+    labs(title = "Total VAT liability of small businesses in 2019 by month",
+         x = "Month",
+         y = "Georgian Lari") +
+    scale_x_continuous(breaks = 201901:201912) +
+    theme(plot.title = element_text(hjust = 0.5))
+
+# Saving
+ggsave("vat_liability_small_2019_text.png",
+       width = 20,
+       height = 10,
+       units = "cm")
+
+# Exercise 4
+  ggplot(small_business_2019_all) +
+    aes(x = age,
+        y = vat_liability) +
+    geom_point() +
+    labs(title = "VAT liability versus age for small businesses in 2019",
+         x = "Age of firm (years)",
+         y = "VAT liability") +
+    theme(plot.title = element_text(hjust = 0.5))
+
+# Saving
+  ggsave("scatter_age_vat.png",
+         width = 20,
+         height = 10,
+         units = "cm")
+

Presentations-GeorgiaRS/202309/libs/kePrint/kePrint.js

@@ -0,0 +1,8 @@
+$(document).ready(function(){
+    if (typeof $('[data-toggle="tooltip"]').tooltip === 'function') {
+        $('[data-toggle="tooltip"]').tooltip();
+    }
+    if ($('[data-toggle="popover"]').popover === 'function') {
+        $('[data-toggle="popover"]').popover();
+    }
+});