merinehlsen
commented
5 months ago Hi Jan,
Thank you so much for your kind feedback! :)
I can definitely add the direct link and the title of the cited papers in the text.
Regarding the picture, you are right, there is supposed to be a picture there. I have been having some trouble inserting it into both the RMarkdown file AND the website. When I am using the absolute path () it is visible on the website but not in Markdown, while  displays it in (my) Markdown file but not on the website. So if you have an idea of how to fix this (besides using the link to the picture, which also did not work...), please let me know! Maybe @martin-raden could also offer some advice? Either way I will be inserting a link for the picture. :)
As for my goals:
You want a combined table, a better visualization (not a bar plot) and a visualization of the effect of dietary composition.
Yes, that is it in a nutshell! I got the feedback from Martin, that the visualization goal is too vague, so I will add some examples of what I am envisioning next week, but you already hit the nail on the head with the linear regression. I am imagining (for example) a plot with protein content on the x axis and TGA content on the y axis and dots with their respective regression lines for the three different diets. I hope that clears it up a bit for now!
Regarding the barplots: There really just a personal preference, but a barplot creates the illusion of data points, meaning that since the bar is draw from 0 to the mean, you visually get the impression that there are values spanning that space, even when there are usually not. Additionally, a barplot generally indicates only the mean of a group (and possibly the Standard deviation or error with the whiskers). A box (and whiskers) plot on the other hand can indicate much more characteristics of how the values are distributed within that group, namely the median/mean, the Interquartile Range and outliers (described also here). So boxplot much better represent the actual range of values and parts of their distribution, which the barplot essential hides. I found this picture online, which makes this point very clear (albeit in a very dramatic way):
That is not to say that barplots are "bad", I'd say often they are the best tool for visualization because they maximise the impression of differences between means and are thus much easier to understand. But in my research I am specifically interested in variance of distributions, so I much prefer boxplots (also often in combination with either the raw data points or a violinplot!). I hope this answers your question!
Again, thank you for your valuable feedback, and wishing you a nice weekend. :)
Best, Meri
Hi Meri,
your project sounds super interesting and relevant. I have a few points I like to add:
Could you include the direct links to the paper and title in the project description itself? You placed it at the buttom but it would be much easier to have a link at the point where you are mentioning your sources.
Did you mean to include the picture for the flies life cycles at section "2. Development time"? As for now, there are just the subtitles for the pictures. I would suggest to include the pictures or to include a link to them.
Just to make I understood your overall goals correctly: You want a combined table, a better visualization (not a bar plot) and a visualization of the effect of dietary composition. I am wondering why the barplot is not sufficient enough and if the effect should be visualized with a linear regression?
I hope these points help you to further develope your project. Have a nice weekend :) Best, Jan