For a context this image was created by revoU to know which companies have the highest rating for happiness index, the data comes from two different source Jobstreet and Glassdoor. A full article can be read in here.
While the data is interesting in term visual it uses incorrect chart and the goal here to suggest a possible way to show the data with different chart, also provide the code so everyone can reproduce it.
What’s the problem with the chart?
The problem appeared when the chart that being used is stacked bar. Index from two different data source should not be merged into single bar. For example Jobstreet happiness rating on Dana vs Mekari it had same value but since Glassdoor rating Dana is higher than Mekari it seems Dana had a higher rating for Jobstreet if we ignore the label.
Above example also means the x-axis for the chart would be double from maximum five into 10.
What could be solution?
Solution one Dumbbell Chart
If the goal is to show two different index then dumbbell chart could be solution. It shows the value for both Jobstreet and Glassdoor and keep the maximum index to five.
Solution two Scatter Plot
From the data we know there are 2 index and also total review from both index, then we can use scatterplot. Use this type of chart, we can also create another insight which companies have value that above average for both Glassdoor and Jobstreet and vice versa.
Conclusion
Different charts can yields different insights. Dumbbell charts are useful to see the rank based on certain metric, however if the data have a lot of observation the chart can be too long. Scatter plot allow user to see how the distribution is and they can make their own conclusions, with help of annotation in this example color code above or below average will help users even more.
All completed code that being used for this chart can be seen in Github
Dumbbell Code
final_plot <- ggplot()+
## start line
geom_segment(data=happiness_index, aes(y=companies, yend=companies, x=2.5, xend=5.2), color="#b2b2b2", size=0.15) +
geom_dumbbell(data = happiness_index, aes(y = companies,x=glassdoor_rating,xend=jobstreet_rating),
colour_x = "#2FC1B9",colour_xend = "#197770",color="#e3e2e1", size=2) +
geom_text(data = filter(happiness_index,companies=="Payfazz"),aes(y = companies,x=glassdoor_rating,label="Glassdoor"),
vjust=-1.5,hjust=1.15,size=3,color="#2FC1B9") +
geom_text(data = filter(happiness_index,companies=="Payfazz"),aes(y = companies,x=jobstreet_rating,label="JobStreet"),
vjust=-1.5,hjust=-0.10,size=3,color="#197770") +
geom_rect(data=happiness_index, aes(xmin=5.1,xmax=5.4,ymin=-Inf,ymax=Inf),fill="grey")+
geom_rect(data=happiness_index, aes(xmin=5.45,xmax=5.75,ymin=-Inf,ymax=Inf),fill="grey")+
geom_text(data=happiness_index, aes(y = companies,x=5.25,label=round(avg_rating,1)),size=3)+
geom_text(data=happiness_index, aes(y = companies,x=5.6,label=total_review),size=3)+
geom_text(data=filter(happiness_index,companies=="Payfazz"), aes(x=5.25,y=companies,label="Avg\nRating"),size=3.2,vjust=-0.52)+
geom_text(data=filter(happiness_index,companies=="Payfazz"), aes(x=5.6,y=companies,label="Total\nReview"),size=3.2,vjust=-.52)+
geom_text(data=happiness_index, aes(y=companies, x = glassdoor_rating, label=glassdoor_rating),vjust=2,size=2.5,color="#2FC1B9")+
geom_text(data=happiness_index, aes(y=companies, x = jobstreet_rating, label=jobstreet_rating),vjust=2,size=2.5,color="#197770")+
scale_x_continuous(expand=c(0,0), limits=c(2.5, 5.8))+
scale_y_discrete(expand = c(0.05,0)) +
theme_bw() +
theme(
panel.grid.major=element_line(colour = "white"),
panel.grid.minor=element_blank(),
panel.border=element_blank(),
axis.ticks=element_blank(),
axis.text.x=element_blank(),
plot.title=element_text(size = 16, face="bold"),
plot.title.position = "plot",
plot.subtitle=element_text(face="italic", size=12, margin=margin(b=12)),
plot.caption=element_text(size=8, margin=margin(t=12), color="#7a7d7e")
) +
labs(title = "Happiness Index Perusahaan Teknologi",
subtitle = "Ranking Perusahaan Teknologi Indonesia dengan
Pendaan Terbesar berdasarkan Rating Karyawan",
x = "Rating",
y=NULL,
caption = "Sumber data dan informasi https://journal.revou.co/happiness-index-perusahaan-teknologi/
Dibuat ulang oleh @data_wizart | Dio Ariadi")
final_plot +
annotate("segment", x = 3.9, xend = 4.25, y = 23, yend = 23, colour = "grey", size=1, arrow=arrow(length = unit(2.5,"mm"),type = "closed"))+
annotate(
geom = "text", x = 3.7, y = 23,label="Glassdoor & Jobstreet\nmemiliki nilai sama",size=2.5)
Scatter Plot Code
scatter <- ggplot()+
geom_rect(aes(xmin = 3.6,xmax = mean(happiness_index$jobstreet_rating),ymin=2.5,ymax=mean(happiness_index$glassdoor_rating)),fill="red",alpha=0.2)+
geom_rect(aes(xmin = mean(happiness_index$jobstreet_rating),xmax = 4.85 ,ymin=2.5,ymax=mean(happiness_index$glassdoor_rating)),fill="yellow",alpha=0.2)+
geom_rect(aes(xmin = 3.6,xmax = mean(happiness_index$jobstreet_rating),ymin=mean(happiness_index$glassdoor_rating),ymax=4.8),fill="yellow",alpha=0.2)+
geom_rect(aes(xmin = mean(happiness_index$jobstreet_rating),xmax = 4.85 ,ymin=mean(happiness_index$glassdoor_rating),ymax=4.8),fill="green",alpha=0.2)+
geom_point(data = happiness_index, aes(x = jobstreet_rating,y = glassdoor_rating,size=(total_review_gd+total_review_js)),
alpha = 0.5,shape=1)+
geom_text_repel(data=happiness_index,aes(x = jobstreet_rating,y = glassdoor_rating,label=companies),size=2.5,force =1,min.segment.length = 0.5)+
geom_hline(yintercept = mean(happiness_index$glassdoor_rating),size=0.2,linetype="dashed")+
geom_vline(xintercept = mean(happiness_index$jobstreet_rating),size=0.2,linetype="dashed")+
# geom_text()
theme_bw() +
theme(
# panel.grid.major=element_line(colour = "grey"),
# panel.grid.minor=element_blank(),
panel.border=element_blank(),
# axis.ticks=element_blank(),
# axis.text.x=element_blank(),
plot.title=element_text(size = 16, face="bold"),
plot.title.position = "plot",
plot.subtitle=element_text(face="italic", size=8, margin=margin(b=12)),
plot.caption=element_text(size=8, margin=margin(t=12), color="#7a7d7e"),
legend.position = "none"
)+
labs(x="Jobstreet Rating",
y = "Glasdoor Rating",
title = "Happiness Index Perusahaan Teknologi di Indonesia",
subtitle = "Berdasarkan rating dari Glassdoor dan Jobstreet
Tiap lingkaran merepresentasikan satu perusahaan. Semakin besar lingkaran semakin banyak jumlah review",
caption = "Sumber data dan informasi https://journal.revou.co/happiness-index-perusahaan-teknologi/
Dibuat ulang oleh @data_wizart | Dio Ariadi
")+
geom_label(aes(x=3.71,y=2.7,label="Di bawah rerata
Jobstreet & Glassdoor"),size=2.3)+
geom_label(aes(x=3.723,y=4.17,label="Di bawah rerata Jobstreet
di atas rerata Glassdoor"),size=2.3)+
geom_label(aes(x=4.73,y=2.7,label="Di bawah rerata Glassdoor
di atas rerata Jobstreet"),size=2.3)+
geom_label(aes(x=4.75,y=4.17,label="Di atas rerata
Glassdoor & Jobstreet"),size=2.3)+
scale_y_continuous(expand = c(0,0),limits = c(2.5,4.8))+
scale_x_continuous(expand = c(0,0),limits = c(3.6,4.85))
scatter