Peer-Review Analysis with R
Budget: €8 – €10,000 EUR
I teach a seminar where students make short presentations in front of the class and everybody rates the presentation on a standardized excel-form (see attachment). The rating is done in freeform text and on a 1-5 scale (-- - 0 + ++) for several categories, such as media use or rhetoric. Students fill out one of these forms for every presentation. So when class-size is 20, there are 20x(20-1) = 380 forms. What I need to extract from these forms is:
- How good was the presentation (compared to all others)
- How good a critic is each student (compared to all others)
Therefore, I need an R-Script (or several) that does the following things:
- Ask for Basic Information: What was the Name of the Seminar, what Semester is it, what is the Import-Folder, what is the Output-Folder, where is the file with Meta-Information about all presentations?
-Import the Meta-Information from a csv-file. In this file, every student (and thus presentation) is assigned a number. Through this number, every excel-form is uniquely connected to a presenter and a reviewer, as every student puts his/her number in as the reviewer and the number of the presenter.
-Import all Excel-Files in the Import-Directory into a Dataframe. Only relevant cells should be kept! Convert the x on the categorical scale (-- - 0 + ++) to the corresponding value on a numeric scale centered on 0 (-2 -1 0 1 2). One row of the data frame should correspond to one excel-File with one column per relevant cell.
- For each student, a PDF report is created (see attachment). The first page contains Meta-Information about the presentation, coming from the Meta-File. Additional scores are displayed, showing how many forms the student filled out and to what degree those forms were complete (%). The second page contains 8 Histogram-Plots, using the numeric feedback on the students own presentation. Each Histogram should contain not only the students presentation result, but also the mean result across all presentations as a reference. Additionally, each plot contains a vertical line for the average rating of the students presentation and all presentations. Page three is built in the same way as page two, but all the ratings the student has given are the basis of the plots, instead of the ratings the student has received. The last pages contain two tables with free-text feedback. The first with feedback for the students presentation, sorted by category. The second with feedback the student has given, sorted in the same way. The whole PDF should have some general formatting with page-numbering, a logo on each page, and the name of the presenter/the seminar.
- Export a csv file with the Meta-Information, but add the average numerical feedback scores for each student and how many forms (and to what degree) each student has filled out.
- How good was the presentation (compared to all others)
- How good a critic is each student (compared to all others)
Therefore, I need an R-Script (or several) that does the following things:
- Ask for Basic Information: What was the Name of the Seminar, what Semester is it, what is the Import-Folder, what is the Output-Folder, where is the file with Meta-Information about all presentations?
-Import the Meta-Information from a csv-file. In this file, every student (and thus presentation) is assigned a number. Through this number, every excel-form is uniquely connected to a presenter and a reviewer, as every student puts his/her number in as the reviewer and the number of the presenter.
-Import all Excel-Files in the Import-Directory into a Dataframe. Only relevant cells should be kept! Convert the x on the categorical scale (-- - 0 + ++) to the corresponding value on a numeric scale centered on 0 (-2 -1 0 1 2). One row of the data frame should correspond to one excel-File with one column per relevant cell.
- For each student, a PDF report is created (see attachment). The first page contains Meta-Information about the presentation, coming from the Meta-File. Additional scores are displayed, showing how many forms the student filled out and to what degree those forms were complete (%). The second page contains 8 Histogram-Plots, using the numeric feedback on the students own presentation. Each Histogram should contain not only the students presentation result, but also the mean result across all presentations as a reference. Additionally, each plot contains a vertical line for the average rating of the students presentation and all presentations. Page three is built in the same way as page two, but all the ratings the student has given are the basis of the plots, instead of the ratings the student has received. The last pages contain two tables with free-text feedback. The first with feedback for the students presentation, sorted by category. The second with feedback the student has given, sorted in the same way. The whole PDF should have some general formatting with page-numbering, a logo on each page, and the name of the presenter/the seminar.
- Export a csv file with the Meta-Information, but add the average numerical feedback scores for each student and how many forms (and to what degree) each student has filled out.