Homework 004 is reading for grading

[mylinhthibodeau]

Dear @KateJohnson,

Excellent homework 4!

Peer review

• Very good annotation of process and examples of challenges to tackle (I have experienced the same challenges actually, so I think they might be universal?)
• Data reshaping: successfully completed Activity 2 - Reshape the gapminder data with spread and gather functions.
• Join, merge, look up: successfully completed Activity 1 - create a new dataframe and join with gapminder data.

I really liked your homework, as it was very tidy, and if I could make one small recommendation, it would be to include some of the "failed attempts" you made to illustrate clearly what has been tried so far to resolve the challenges, which can be helpful to generate hypotheses.

Additional information

1. I had the same problem than you trying to label groups on the spread version of my data, but I concluded that the most efficient way was to use the long version of the table.

For example, in your case, this code would label the country appropriately in a different colour and the legend would accompany your graph:

ggplot(gap.long, aes(x=year, y= lifeExp, colour = country)) + 
  geom_point() +
      xlab("Year") + ylab("Life Expectancy") + theme_bw() 

If you found out how to label by "columns" (country) using your gap.spread table format, please do let me know !!

2. I tried to look into how to match columns according to a substring, but did not succeed.

I was able to solve the "Congo" problem using the fuzzyjoin package, as exemplified here, because "Congo" (conco dataset) is a substring of "Congo, Dem. Rep." (gap.le dataset)

library(fuzzyjoin)
conco_gap.le <- gap.le %>% regex_left_join(conco, by = "country")
View(conco_gap.le)
dim(conco_gap.le)

However, it created a new problem: "United States of America" (conco dataset) is not a substring of "United States" (gap.le dataset), it did not appropriately match for United States :( Also, Guinea and Equatorial Guinea had two different iso.3c and it duplicated the data entries for Guinea in the gap.le table.

Once again, if you do find the solution for this, please pretty please let me know, because I would really like to know how to solve this problem.

Warm regards, My Linh Thibodeau

[abishekarun]

Peer Review:

Hi, @KateJohnson ! You did an excellent homework and went ahead and explored different dataset(country code).

Data reshaping

• It was great to see that you had performed the given task using gather() and spread() functions.
• It was also nice to see that you had mentioned the challenges that you had faced.
• Even I would like to know how ggplot2 handles wide data. I have heard that long version of data is more suitable for ggplot2. When I looked into possible solutions for the problem that you had faced, I couldnt anything without reshaping the data. So it would be nice to know if this can be done.

Data joining

• All the join functions were explored properly with the two datasets that you had chose.
• I also would like to know the solution for matching the columns. I tried using matching functions to get the common rows between two datasets.I was able to perform anti_join and semi_join but couldnt do the mutating joins. It would be nice to know how this can be done. I used str_detect() function to match the two strings. Kindly let me know if you find out how this can be done.

Some suggestions

• One possible suggestion is to look into suppressWarnings(suppressMessages()) for removing the warning messages. It just makes the markdown file more clean.
• One suggestion would be to proofread and possibly play around with kablestyling option and pandoc options for formatting the tables.
• One small suggestion would be to put the link to the markdown file in the issue as it will be easier to access it.

The fact that you had mentioned the struggles that you had experienced is very helpful and useful for other students. It was also nice to see you mention the resources that helped you.

Overall, I think your homework was really well done and hope you can keep it up!

Regards, Arun Rajendran

[derekcho]

Hi @KateJohnson, here are some comments about your hw04:

• Nice work successfully converting your data to wide form with the spread function. The scatterplot that you showed is done much more easily with long form data (as I’m sure you noticed for yourself). An example of a plot that is easier with wide form data is where you have the life expectancies of one country on the y-axis and the life expectancies of another country on the x-axis.
• In your second task you applied a left join to add country codes to a subset of the gapminder data. You provided a good explanation as to why NAs were introduced when values of country were present in one data frame but not the other. However, it would have been nice if you tried to explore the effects of other joins as well rather than focusing on only left joins.
• Your report was well organized and formatted. In general, your code is efficient and easy to read. Your repo is well organized. Good work!

Your grade will be emailed to you at a later date.