In this report, we use data from the famous http://www.gapminder.org website. Let’s look at some random rows of this dataframe:
library(gapminder)
data(gapminder)
gapminder %>% sample_n(10) %>% kable| country | continent | year | lifeExp | pop | gdpPercap |
|---|---|---|---|---|---|
| Central African Republic | Africa | 1967 | 41 | 1733638 | 1136 |
| Benin | Africa | 1987 | 52 | 4243788 | 1226 |
| Algeria | Africa | 1952 | 43 | 9279525 | 2449 |
| Egypt | Africa | 2007 | 71 | 80264543 | 5581 |
| Oman | Asia | 1997 | 72 | 2283635 | 19702 |
| Uruguay | Americas | 2007 | 76 | 3447496 | 10611 |
| Singapore | Asia | 2002 | 79 | 4197776 | 36023 |
| Gabon | Africa | 2007 | 57 | 1454867 | 13206 |
| Tunisia | Africa | 1962 | 50 | 4286552 | 1660 |
| Ecuador | Americas | 1962 | 55 | 4681707 | 4086 |
Here we see how many different values we have per column:
df1 <- gapminder %>% sapply(n_distinct) %>% t %>% t
data.frame(Variable = row.names(df1), Values = df1[,1],row.names = NULL) %>% kable| Variable | Values |
|---|---|
| country | 142 |
| continent | 5 |
| year | 12 |
| lifeExp | 1626 |
| pop | 1704 |
| gdpPercap | 1704 |
Here is how many countries we have per continent:
gapminder %>% group_by(continent) %>% summarise("Number of Countries" = n_distinct(country)) %>% kable| continent | Number of Countries |
|---|---|
| Africa | 52 |
| Americas | 25 |
| Asia | 33 |
| Europe | 30 |
| Oceania | 2 |
In Oceania, we only have Australia and New Zealand.
First, we are going to look at the Evolution of Total Population per continent:
library(highcharter)
PG <- gapminder %>% group_by(continent, year) %>% summarise("pop" = sum(as.numeric(pop)))
hchart(PG, "line", hcaes(x = year, y = pop, group = continent)) %>%
hc_tooltip(shared=TRUE, table=TRUE)# library(plotly)
# p <- ggplot(PG, aes(x=year, y=pop, color=continent)) +
# geom_line(size=1)
# ggplotly(p)The population growth in Asia is striking! Africa’s recent growth and Europe’s stagnation is also visible.
We can also display the stacked values in order to see the total population:
PG$continent <- factor(PG$continent,levels = rev(c("Oceania","Europe","Americas","Africa","Asia")))
hchart(PG,"area",hcaes(x=year,y=pop,group=continent)) %>%
hc_tooltip(shared=TRUE, table=TRUE,crosshairs=TRUE) %>%
hc_plotOptions(area= list(stacking="normal"))# library(streamgraph)
# streamgraph(PG, "continent", "pop", "year", offset="zero") %>%
# sg_fill_brewer("Blues")First, let’s look at the life expectancy. The ribbons of the following figure represent the 25- and 75-percentile for each continent over time.
LE <- gapminder %>% group_by(continent, year) %>% summarise(LE = mean(lifeExp),
low = quantile(lifeExp,0.25),
high = quantile(lifeExp,0.75))
hchart(LE,"arearange", hcaes(group=continent)) %>%
hc_xAxis(categories = LE$year) %>%
hc_tooltip(shared=TRUE,crosshairs=TRUE)# ggplot(LE, aes(x=year, y=LE, fill=continent)) +
# # geom_line(aes(color=continent), size=1) +
# geom_line(data=gapminder, aes(x=year,y=lifeExp,group=country,color=continent), size=1.5, alpha=.2) +
# geom_ribbon(aes(ymin=low, ymax=high), alpha=0.8) +
# scale_fill_brewer(palette = 'Spectral') +
# scale_color_brewer(palette = 'Spectral')What is that outlier? In 1992, Rwanda hat a Life Expectancy of 23.6 years …
Note that the ribbon for Oceania is very narrow because there are only 2 Countries in this category. We notice that LE in Asia increased a lot, while LE in Africa started decreasing in the mid 1980s.
Now we are going to investigate the GDP per capita
# remove Kuwait because very odd values prior to 1970s
gapminder %>% filter(!(country %in% "Kuwait")) %>%
hchart("line",hcaes(x=year,y=gdpPercap,group=country, color=continent)) %>%
hc_legend(enabled=FALSE) %>%
hc_yAxis(title=list(text="GDP per capita"))Too crowded, let’s look at the grouped data:
GDP <- gapminder %>% group_by(continent, year) %>% summarise(gdp = mean(gdpPercap),
low = quantile(gdpPercap,0.25) %>% round(),
high = quantile(gdpPercap,0.75) %>% round())
hchart(GDP,"arearange", hcaes(group=continent)) %>%
hc_xAxis(categories = GDP$year) %>%
hc_tooltip(shared=TRUE,crosshairs=TRUE)From 1987 onwards, especially Europe and Asia are very large. Let’s cut these groups in two.
GDP1 <- gapminder %>% filter(year>1985) %>% filter(continent %in% c("Asia","Europe")) %>%
group_by(continent,country) %>% summarize(avg=mean(gdpPercap)) %>% arrange(avg) %>% mutate(g=ifelse(avg<1e4,"low","high")) %>% ungroup() %>% mutate(group=paste(continent,g))
g4 <- GDP1 %>% group_by(group) %>% select(country) %>% ungroup()
g4$country <- as.character(g4$country)
x <- list()
for(i in unique(g4$group)) {
x[[i]] <- g4 %>% filter(group==i) %>% select(country) %>% distinct(country) %>% pull
}
x## $`Asia low`
## [1] "Myanmar" "Afghanistan" "Cambodia"
## [4] "Nepal" "Bangladesh" "Vietnam"
## [7] "India" "Pakistan" "Yemen, Rep."
## [10] "Mongolia" "Korea, Dem. Rep." "Philippines"
## [13] "China" "Indonesia" "Sri Lanka"
## [16] "Syria" "Jordan" "West Bank and Gaza"
## [19] "Thailand" "Iraq" "Lebanon"
## [22] "Iran" "Malaysia"
##
## $`Europe low`
## [1] "Albania" "Bosnia and Herzegovina"
## [3] "Turkey" "Bulgaria"
## [5] "Montenegro" "Romania"
##
## $`Europe high`
## [1] "Serbia" "Poland" "Croatia"
## [4] "Slovak Republic" "Hungary" "Czech Republic"
## [7] "Portugal" "Slovenia" "Greece"
## [10] "Spain" "Italy" "Finland"
## [13] "Ireland" "France" "United Kingdom"
## [16] "Sweden" "Belgium" "Germany"
## [19] "Iceland" "Austria" "Denmark"
## [22] "Netherlands" "Switzerland" "Norway"
##
## $`Asia high`
## [1] "Korea, Rep." "Taiwan" "Oman"
## [4] "Israel" "Saudi Arabia" "Bahrain"
## [7] "Japan" "Hong Kong, China" "Singapore"
## [10] "Kuwait"g1 <- GDP1 %>% filter(g=="low") %>% select(country,group) %>% inner_join(gapminder) %>%
group_by(group, year) %>% summarise(gdp = mean(gdpPercap),
low = quantile(gdpPercap,0.25) %>% round(),
high = quantile(gdpPercap,0.75) %>% round())
g2 <- GDP1 %>% filter(g=="high") %>% select(country,group) %>% inner_join(gapminder) %>%
group_by(group, year) %>% summarise(gdp = mean(gdpPercap),
low = quantile(gdpPercap,0.25) %>% round(),
high = quantile(gdpPercap,0.75) %>% round())
g3 <- bind_rows(g1,g2)
g3 %>% ungroup() %>%
hchart("arearange",hcaes(group=group)) %>%
hc_xAxis(categories = g3$year)We can clearly see that there are two categories of countries in both continents Europe and Asia. It’s striking that the GDP per capita in the Europe low group fell between 1987 and 1992. Might this have something to do with the fall of the communist regimes?
Now, let’s investigate some countries which suffered a decrease in GDP per capita. Between 1987 - 1992, 43% of the countries suffered a decrease in GDP per capita.
gdp <- gapminder %>% filter(year %in% c(1987,1992))
gdp <- gdp %>%
group_by(continent,country) %>%
summarize(diff=100*(last(gdpPercap)-first(gdpPercap))/first(gdpPercap)) %>%
arrange(diff)
hchart(gdp[1:20,], "column", hcaes(x=country,y=round(-diff))) %>%
hc_yAxis(title= list(text="GDP decrease in %"))Especially Iraq has been hit very strongly (see Gulf War 1990/91)
The following figure shows the 10 countries with the worst GDP performance between 1987 and 1992:
gdp2 <- gdp[1:10,] %>% ungroup() %>% select(country) %>%
inner_join(gapminder) %>% filter(year>1986)
hchart(gdp2,"line",hcaes(x=year,y=gdpPercap,group=country))Here, we are going to focus on the 26 African and 13 Asian countries which had a GDP per capita below 1000 USD in the 1950s.
gdp3 <- gapminder %>% filter(year %in% c(1952,1957)) %>% group_by(country) %>%
summarise(avg=mean(gdpPercap)) %>% arrange(avg)
# gdp3$avg[1:39]
# gdp3[1:39,"country"] %>% inner_join(gapminder %>% filter(year==1987) %>% select(country,continent)) %>% select(continent) %>% table
gdp3[1:39,"country"] %>% inner_join(gapminder) %>%
hchart("line",hcaes(x=year,y=gdpPercap,group=country)) %>%
hc_legend(enabled=FALSE)Now, some of them had an impressive development! Up to you to check the individual cases and find the reasons behind these promising examples.
library(plotly)
p <- ggplot(gapminder, aes(gdpPercap, lifeExp, color = continent)) +
geom_point(aes(size = pop, frame = year, ids = country)) +
scale_x_log10()
ggplotly(p) %>% animation_opts(1000, easing = "bounce", redraw = FALSE)