Assignment 2

Deconstruct, Reconstruct Web Report

Original

Source: BMJ Journal Public health Research (2014).

Objective

The objective of the graph is to determine trend in the years of potential life lost in india with respect to sex and the most frequent diseases that caused the death. The main audience of this data visualisation are people of health sectors, medical researcch students and also common man whose knowledge about the data and visualisations could be limited.

The visualisation chosen had the following three main issues:

• Deceptivity: It is unavoidable to not notice the X axis of the graph. The visualisation doesn’t follow the conventional presentation of the data. The number of deaths are represented on the x axis which makes it difficult to interpret.Also the X and Y axis labels are missing.The horizontal dual faced graph makes it hard to compare the change in the trend of deaths for various diseases between male and female.The visualisation is unaesthetic and deceptive.

• Inability to satisfy the objective: Through this graph, it is highly time consuming to visualise and interpret the trend in deaths caused by different diseases. The graph talks about the number of deaths rather than the rate of deaths. Rate is a better measure to analyse than the number of deaths as the size of the total population for each disease is unknown and may vary. Thus the visualisation failed to answer the objective. Also,the data presented is from a collection of different sources of data which rise questions on the reliability and integrity.

• Improper colour choice: The colours blue and red are usually associated with tritanopia - unable to perceive blue and tritanomaly - reduced sensitivity to blue . Though the percent of people with this disorder is comparatively very low, it still reflect a bad visualisation practice. Also, the graph becomes tricky in the grey scale with the dark shades used. For instance, if this visualisation is printed in an article in the newspaper or a book, it is hard to comprehend the point of contact for the 2 legends- male and female.

Reference

Dubey,M & Mohanty,S.K.(2014).Age and sex patterns of premature mortality in India:Figure 1. BMJ Open, 4(8). https://bmjopen.bmj.com/content/4/8/e005386

Code

The following code was used to fix the issues identified in the original.

library(readxl)
library(dplyr)
library(ggplot2)
library(reshape2)

vis_data <- read_excel("C:/Users/intel/Desktop/vis data.xlsx")
new <- vis_data %>% subset( ,c(1,2,3))
new2 <- new[-c(1), ]

names(new2) <- lapply(new2[1, ], as.character)
df <- new2[-1,] 
head(df)

## # A tibble: 6 x 3
##   Causes_of_death     Y.Male             Y.Female          
##   <chr>               <chr>              <chr>             
## 1 Tuberculosis        16.7               13                
## 2 Diarrhoeal          21.4               17.600000000000001
## 3 Malaria             7.9                7.8               
## 4 Malignant,neoplasms 11.9               14.6              
## 5 Cardiovascular      38.200000000000003 33.299999999999997
## 6 Respiratory         15.5               16

df$Y.Male <- as.numeric(as.character(df$Y.Male))
df$Y.Female <- as.numeric(as.character(df$Y.Female))
head(df)

## # A tibble: 6 x 3
##   Causes_of_death     Y.Male Y.Female
##   <chr>                <dbl>    <dbl>
## 1 Tuberculosis          16.7     13  
## 2 Diarrhoeal            21.4     17.6
## 3 Malaria                7.9      7.8
## 4 Malignant,neoplasms   11.9     14.6
## 5 Cardiovascular        38.2     33.3
## 6 Respiratory           15.5     16

Rate_of_deaths <- melt(df, id.vars = "Causes_of_death", measure.vars = c("Y.Male","Y.Female"))
head(Rate_of_deaths)

##       Causes_of_death variable value
## 1        Tuberculosis   Y.Male  16.7
## 2          Diarrhoeal   Y.Male  21.4
## 3             Malaria   Y.Male   7.9
## 4 Malignant,neoplasms   Y.Male  11.9
## 5      Cardiovascular   Y.Male  38.2
## 6         Respiratory   Y.Male  15.5

p<-ggplot(data=Rate_of_deaths, aes(x=Causes_of_death, y= value, fill= variable),cex.axis = 0.1, width = "100%") +
  geom_bar(stat="identity",color= "Black", position=position_dodge()) +
  scale_x_discrete(guide = guide_axis(n.dodge=2))+
  theme_classic() 

p1 <- p + scale_fill_manual(values = c("Cyan4","Pink"))+
  labs(title = "Rate Of YPLL deaths for most frequent types of deaths in India",
              subtitle = "With respect to gender",
              caption = "YPLL = Years of Potential Life Lost" ,
              x = "Causes of death", y = "Rate of deaths per 1000 people")
ggsave("Rate of YPLL deaths in people of India.png", width=25)

Data Reference

Dubey,M & Mohanty,S.K.(2014).Age and sex patterns of premature mortality in India:Table 1. BMJ Open, 4(8). https://bmjopen.bmj.com/content/4/8/e005386

Reconstruction

The following plot fixes the main issues in the original.