R Notebook

Deconstruct and Reconstruct

Student Details

Wesley Paul Nderi (s3635870)

Part I - Choosing a Visualisation

Taken from: Plotly Community Feed

Part II - Deconstruct

Borrowing from Kaiser Fung’s Trifecta check up to guide our critique:

The above visualisation tackles the question of whether there exists a relationship between a father’s job and the level of education reached by a candidate. This is certainly an interesting question. Although the data sourced is a summary and not cite what country or city this information, it is still sufficient to bring out some interesting trends.

The visual does an okay job to convey that there is a difference in the levels of education of a candidate based on the job of the father. We can definitely see a relationship between the two, for example, there are much fewer number of graduates in the Daily Labour category than there within the Government Job category. This said, we see that in general there are very limited number of Post graduates in each category. Does this mean that this particular subset is unaffected by the category of the father’s job? Could it also possibly a limited amount of data in that category?

On the flip side, there is also a clear domination of the visualisation by the Graduate and 12th class categories.Could it also possibly an overrepresentation of the amount of data in that category?

Another issue is the y axis labels. The Educational Level of the Candidate is put on the y axis but looking at the data, it appears that it is supposed to be a count. This is a miscommunication as it can deceive the user that the y axis is being used as a measurement scale for the level of education.

Part III - Reconstruct

Formatting the data

felc2<-melt(felc)

colnames(felc2) = c("FathersJob", "Grade", "Number")

felc2$Grade<- felc2$Grade %>%factor(levels=c("X.10th.Class..y.","X.12th.Class..y.","X.Graduate..y.","X.Post.Graduate..y."), labels=c("10th Class","12th Class", "Graduate", "Postgraduate"), ordered=TRUE)

ggplot(data= felc2, aes(x=FathersJob, y=Number)) + geom_point(aes(color=Grade)) + scale_y_log10() + theme_bw() + geom_boxplot(fill=NA) + geom_jitter(alpha=0.1) + ggtitle("How does a father's job impact the level of education reached?") + labs(x= "Father's Job Category", y="Number of Students") 

The above visualisation is now able to depict the story behind the data. It is also simple and easy to read.We see that although the highest level of education reached in Farming category was the 12th Class, it has the highest median in all of the categories whereas comparatively although the Government Job category has both Graduates and Post graduate students it has the lowest median. This means that the former category has more students going to school overall.

We are also able to see interesting trends in variation(reading from the respective IQRs), for example, we notice that there is a lot of variation in Owned Business category between the number of students reaching the 12th class and the number of students getting to Postgraduate level which is something that could be inspected. It could also possibly be a indicator of the quality of the data.

It is also easier to note that in each of the categories, there are very few students who reach the Post graduate level and note that interestingly enough, it is lowest in the Government Job category.

In addition, we are now able to see an outlier in the Farming category which is again may speak of the quality of the data or call for some introspection of the category.

References

1.Plotly Community Feed