Gapminder

Observations for life expectency since 1952

The life expectency has been increasing linearly in most of the countires except a few countires in Africa and Asia. The constant improvement of life expectency can be attributed to access to better health care. Another factor could be the extensive research and development for prevention of diseases. The drop in the life expectency in Africa could be because of the outbreak of HIV-AIDs around mid 1970s. The same might be the cause of drop in a few countries in Asia. Let’s join a few dataframes with more data than the ‘gapminder’ package. Specifically, we will look at data on

• Life expectancy at birth (life_expectancy_years.csv)
• GDP per capita in constant 2010 US$ (https://data.worldbank.org/indicator/NY.GDP.PCAP.KD)
• Female fertility: The number of babies per woman (https://data.worldbank.org/indicator/SP.DYN.TFRT.IN)
• Primary school enrollment as % of children attending primary school (https://data.worldbank.org/indicator/SE.PRM.NENR)
• Mortality rate, for under 5, per 1000 live births (https://data.worldbank.org/indicator/SH.DYN.MORT)
• HIV prevalence (adults_with_hiv_percent_age_15_49.csv): The estimated number of people living with HIV per 100 population of age group 15-49.

We have to use the wbstats package to download data from the World Bank. The relevant World Bank indicators are SP.DYN.TFRT.IN, SE.PRM.NENR, NY.GDP.PCAP.KD, and SH.DYN.MORT

# load gapminder HIV data
hiv <- read_csv(here::here("data","adults_with_hiv_percent_age_15_49.csv"))
life_expectancy <- read_csv(here::here("data","life_expectancy_years.csv"))
# get World bank data using wbstats
indicators <- c("SP.DYN.TFRT.IN","SE.PRM.NENR", "SH.DYN.MORT", "NY.GDP.PCAP.KD")
library(wbstats)
worldbank_data <- wb_data(country="countries_only", #countries only- no aggregates like Latin America, Europe, etc.
                          indicator = indicators, 
                          start_date = 1960, 
                          end_date = 2016)
# get a dataframe of information regarding countries, indicators, sources, regions, indicator topics, lending types, income levels,  from the World Bank API 
countries <-  wbstats::wb_cachelist$countries

We have to join the 3 dataframes (life_expectancy, worldbank_data, and HIV) into one. We may need to tidy your data first and then perform join operations.

1. What is the relationship between HIV prevalence and life expectancy? We generate a scatterplot with a smoothing line to report our results.

skim(hiv)

Variable type: character

Variable type: logical

Variable type: numeric

life_expectency_long <- life_expectancy %>%
  pivot_longer(2:302, values_to = "life_expec" , names_to = "year")
hiv_long <- hiv %>%
  pivot_longer(2:34, values_to = "prop_hiv" , names_to = "year")
life_hiv <- right_join(hiv_long, life_expectency_long, by = c("year" = "year", "country" = "country"))
life_hiv$year = as.numeric(life_hiv$year)
life_hiv_bank_data <- left_join(life_hiv, worldbank_data, by = c("year" = "date", "country" = "country"))
life_hiv_bank_data <- life_hiv_bank_data %>%
  rename(fertilityRate = SP.DYN.TFRT.IN, gdp = NY.GDP.PCAP.KD, mortalityRate = SH.DYN.MORT, primarySchoolEnrolment = SE.PRM.NENR)

We used a mixture of both right and left joins because so long as the dataframes are in the correct order when passed through the function, it does not matter whether you do a right or left join. We did not use an inner join as that would have resulted in data loss - the HIV dataframe only has values from the year 1979 onwards whereas the life expectancy dataframe has values from the year 1800 onwards.

life_hiv_bank_data_region <- right_join( life_hiv_bank_data ,countries, by = c( "country" = "country")) %>%
  select(year, country, region,fertilityRate, gdp, mortalityRate, primarySchoolEnrolment, prop_hiv,life_expec )

1. What is the relationship between HIV prevalence and life expectancy? We generate a scatterplot with a smoothing line to report your results. You may find faceting useful

ggplot(life_hiv_bank_data_region, aes(x=prop_hiv, y=life_expec )) +
  geom_point() +
  scale_y_log10() +
  scale_x_log10() +
  ylim(30,100) +
  geom_smooth(method = "lm", formula= y~x) +
  labs( title = "Relationship HIV prevalence and life expectancy ", y = "Life Expectency", x = "Proportion of HIV prevalance ")

As we can see from the above plot, as the proportion of HIV prevalence increases, the life expectancy decreases. Thus, there is a negative relationship between the two variables.

1. What is the relationship between fertility rate and GDP per capita? We generate a scatterplot with a smoothing line to report our results.

ggplot(life_hiv_bank_data_region, aes(x=fertilityRate , y=gdp )) +
  geom_point() +
  scale_y_log10() +
  scale_x_log10() +
#  ylim(30,100) +
  geom_smooth(method = "lm", formula= y~x) +
  facet_wrap(~region) +
  labs( title = "Relationship fertility rate and GDP per capita ", y = "GDP per capita", x = "Fertility rate ")

As GDP increases, the fertility rate decreases.

1. Which regions have the most observations with missing HIV data? We generate a bar chart (geom_col()), in descending order.

NA_count <- life_hiv_bank_data_region %>%
  group_by(region) %>%
  filter(is.na(prop_hiv)) %>%
  summarise(countNA = n())

ggplot(NA_count,aes( y = fct_reorder(region, countNA), x = countNA)) +
  geom_col()

Sub-Saharan Africa and Europe & Central Asia have the most missing observations for HIV data.

1. How has mortality rate for under 5 changed by region? In each region, we find the top 5 countries that have seen the greatest improvement, as well as those 5 countries where mortality rates have had the least improvement or even deterioration.

mortality_diff <- life_hiv_bank_data_region %>%
  filter(year == 1800 | year == 2021 ) %>%
  group_by(year) %>%
  mutate( region = region, 
          country = country, 
          life_expec = life_expec) %>%
  select(region, country, life_expec) %>%
  pivot_wider( names_from = year, values_from = life_expec) %>%
  rename("year_1800" = "1800", "year_2021"= "2021") %>%
  mutate(change_mortality = (year_2021 - year_1800) )

mortality_top5 <- mortality_diff %>%
  group_by(region) %>%
  slice_max(order_by = change_mortality, n=5)
mortality_top5

## # A tibble: 32 x 5
## # Groups:   region [7]
##    region                country         year_1800 year_2021 change_mortality
##    <chr>                 <chr>               <dbl>     <dbl>            <dbl>
##  1 East Asia & Pacific   Singapore            29.1      85.4             56.3
##  2 East Asia & Pacific   Australia            34        83               49  
##  3 East Asia & Pacific   Thailand             30.4      79               48.6
##  4 East Asia & Pacific   Japan                36.4      84.8             48.4
##  5 East Asia & Pacific   New Zealand          34        82.2             48.2
##  6 Europe & Central Asia Italy                29.7      83.8             54.1
##  7 Europe & Central Asia Spain                29.5      83.6             54.1
##  8 Europe & Central Asia Sweden               32.2      83.1             50.9
##  9 Europe & Central Asia Kyrgyz Republic      23.9      73.3             49.4
## 10 Europe & Central Asia France               34        83.3             49.3
## # ... with 22 more rows

mortality_least5 <- mortality_diff %>%
  group_by(region) %>%
  slice_max(order_by = -change_mortality, n=5)
mortality_least5

## # A tibble: 32 x 5
## # Groups:   region [7]
##    region                country          year_1800 year_2021 change_mortality
##    <chr>                 <chr>                <dbl>     <dbl>            <dbl>
##  1 East Asia & Pacific   Papua New Guinea      31.5      59.3             27.8
##  2 East Asia & Pacific   Cambodia              35        70.9             35.9
##  3 East Asia & Pacific   Mongolia              31.8      69.7             37.9
##  4 East Asia & Pacific   Kiribati              24.9      63.3             38.4
##  5 East Asia & Pacific   Myanmar               30.8      69.5             38.7
##  6 Europe & Central Asia Ukraine               36.6      71               34.4
##  7 Europe & Central Asia Belarus               36.2      74.8             38.6
##  8 Europe & Central Asia Bulgaria              35.8      75.4             39.6
##  9 Europe & Central Asia Romania               35.7      75.7             40  
## 10 Europe & Central Asia Moldova               33.1      73.3             40.2
## # ... with 22 more rows

We can see that each region has seen various changes in mortality rate over the years.

1. Is there a relationship between primary school enrollment and fertility rate?

ggplot(life_hiv_bank_data_region, aes(x=primarySchoolEnrolment, y=fertilityRate)) +
  geom_point() +
  scale_x_log10() +
  scale_y_log10() + 
  geom_smooth(method = "lm", formula = y~x)

 life_hiv_bank_data_region %>%
  select(fertilityRate, primarySchoolEnrolment) %>%
  na.omit() %>%
  cor()

##                        fertilityRate primarySchoolEnrolment
## fertilityRate              1.0000000             -0.7265111
## primarySchoolEnrolment    -0.7265111              1.0000000

As we can see from the scatter plot, there is a strong negative relationship between primary school enrollment and fertility rate. We can see that as primary school enrollment increases, the fertility rate decreases. This is further supported by the correlation between the two variables. The correlation in question is -0.727 which is quite significant negative correlation.