Mortality Rates by Year

0. The World Health Organization publishes data sets which track mortality rates by age group for the years 2013, 2012, 2000 and 1990. This data set is further catagorized by mortality criteria with a 3 character designation. This analysis uses “nqx” or the probability of dying within an age group bracket from the global data set provided.

http://apps.who.int/gho/data/view.main.LIFEWBGLOBAL?lang=en

The World Health Organization Characterizes the raw data used in this analysis as “Mortality and global health estimates”" from the Life tables by World Bank Income Group Global.

This rudimentary analysis looks at the 22 age group brackets for each of the 4 years considered in the WHO’s super-set.
A historgram of mortality and a boxplot contingencies matrix of age bracket to mortality faceted by year is shown.

1. Load Data Frame from file

l <- read.csv(
              "/Users/scottkarr/IS607Spring2016/project2/more/mortality.csv",
              sep=",",
              na.strings = "",
              blank.lines.skip = TRUE,
              col.names = c("Indicator", "Age Group", 
                                "Both Sexes", "Female","Male",
                                "Both Sexes", "Female","Male",
                                "Both Sexes", "Female","Male",
                                "Both Sexes", "Female","Male"
              ),
              stringsAsFactors=FALSE, 
              header=FALSE
    )
df = data.frame(l)
# rows 1 & 2 contain header names
names(df) <- c('Indicator', 'Age.Group',paste(df[1,3:14], df[2,3:14]))
# remove rows 1 & 2
df <- df[-c(1,2),]
names(df) <- gsub("Both sexes:", "", names(df))
#kable(head(df), align = 'l')

2. Tidy data and analyze–let’s look probability of mortality by age bracket

df <- df %>% 
  select(Indicator, Age.Group, ends_with("sexes"))
# return only rows with mortality range indicator nqx
df <- df %>%  
  filter(grepl("nqx", Indicator))
# no longer need Indicator
df <- df %>% 
  select(Age.Group, ends_with("sexes"))
# gather morphs data from wide to long format
df_tidy <- df %>% 
  gather(year, rates, -Age.Group) %>%
  arrange(Age.Group)
# separate year into year and gender 
df_tidy <- df_tidy %>%
  separate(year, into = c("year", "gender"), sep = "\\s") 

## Warning: Too many values at 88 locations: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
## 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, ...

# dispense with gender as a category, we are looking at combined gender mortality
df_tidy <- df_tidy %>% 
  select(Age.Group, year, rates)
# make year and rates numeric
df_tidy$year <- as.numeric(df_tidy$year)
df_tidy$rates <- as.numeric(df_tidy$rates)
# make Age.Group 100+ years consistent so it sorts correctly
df_tidy[13:16,1] <- rep('100-100+',4)
# present data nicely
kable(head(df_tidy), align = 'l')

Age.Group	year	rates
<1 year	2013	0.034089
<1 year	2012	0.035079
<1 year	2000	0.053165
<1 year	1990	0.063140
1-4 years	2013	0.012828
1-4 years	2012	0.013414

3. Presentation

hist(df_tidy$rates)

boxplot(
      df_tidy$rates ~  df_tidy$Age.Group,
        xlab="Increasing Age Group",
        main="Mortality Rates 4 years"
)

bp <- ggplot(data=df_tidy, aes(x=Age.Group, y=rates)) +
      geom_boxplot() +
      geom_text(aes(label = Age.Group, x = Age.Group, y = rates), size=.1, vjust=1, hjust=0.5)

# Split in horizontal direction
bp + facet_grid(. ~ year)