hw06- TA comment

[sherrie9]

• Hw page very organized, thank you for the task bar on the left. Super helpful!
• For the functions: generally you did a good job. For the model_fit() function there is a typo, your wrote typeof(year) however year is not present in the arguments, do you mean typeof(yr)? Your function uses lm fit on variable year. Could you make it more general, so that it could fit a model of any two user supplied variables. (Hint: the first step could be to check if this variable is in gapminder and if it is, we can select that variable for model fitting).
• For plot_model_fit() function, it would be good to have type of models as an argument. You could have this argument taking several choice: linear model, quantile, least square, or all"
• plot choice function: some arguments that are not essential for plotting such as jitter, xlab ylab, etc. should have a default value, so that it doesn't cause an error if user just want to check a plot without putting in all the efforts into giving titles. alpha, hajust should be argument, you set these values inside the function, this is not very dynamic, what if users want to change to a different alpha? Rule of thumb is, unless you are absolutely sure about a value that is not supposed to set by the user, function body should contain as few static numbers as possible.
• List part is good. I like the word count plot!

Overall, good job on this hw!

[rasiimwe]

Hi @sherrie9 Thanks for your kind and helpful comments.
Compared to my previous assignments, this was by far one of my best and most creative but I noticed that I got way less points (88%) - not sure whether this is the new grading style that Dr.Coia mentioned in class. I opted for html because of the plotly function calls that I made in my function myplot(). My tables herein were rendered using kable() - could this be what cost me all those points? Kindly throw more light this.

Many thanks.

[sherrie9]

Hi Rebecca: Html is totally fine. The marking scheme is still the same but the material in 547 is more advanced than previous. Writing good and transferable functions could be tricky. Your tables look fine, but having title would be more helpful. I could have confused someone else's table with yours because I remembered one of tables having too many columns, but I didn't see it in yours after checking it again. I will adjust this mark. And sorry for the confusion!

One more note on functions: the output of the function inspect() is too confusing to me. I appreciate the idea of using one function to do some preprocessing or checking, but it could be better if you pinpoint the output.

Overall, you did put an above and beyond effort into this hw. Let me know if you have other concerns.

[rasiimwe]

Hi @sherrie9

Thanks for your kind and timely feedback!

I wrote the inspect() fuction as a supplementary task. True - when we use list() to produce multiple outputs from a function, it doesn't provide an option or I didn't see one on separating the output so it lumps the output and you need to use indexing to follow the output. The idea was to avoid calling the singular functions class(), str(), head() all the time and I wanted to have a function that runs all these calls for me at once. I will look for another way of representing the output. I really thought I had nailed this assignment with 200% 😄 because I put in alot of effort and did all that was required and more. May be for future work, I should cut down on suplementary tasks/effort to avoid being dissapointed but then, I will miss out on points for doing more 🤔. Thanks for the tips and ideas pointed out.

[sherrie9]

You could use $ operator to select the elements in a list by its name (given you have named the elements in your list). In this way, you could selectively show some output. For example, output$dim could be used to show dimensions. or output$summary etc.. I understand it would be hard to juggle the extra stuff. I do encourage you trying out new things. You did an excellent job. Just to keep in mind of the basics, such as making good plots with proper label names, titles, ratios; illustrative tables with titles. These small things can be quite important but they are generally easy to fix. I noticed that you actually did try to change the x y label names but you used the variable names. lifeExp would not be as a good label name as Life Expectancy.

[rasiimwe]

Oh wow! This is awesome 👍 !! Thanks @sherrie9! I hadn't thought of this perspective. Thanks 💃