Placing incoming students in Mathematics, Statistics, and Computer Science

As I mentioned about two weeks ago, one of my responsibilities is to place incoming students in Mathematics, Statistics, and Computer Science. Why, you may ask, does someone in CS do the placement? The answer is both simple and complex.

It used to be that the department of Mathematics and Computer Science manually placed each incoming student by reviewing their information (e.g., scores on standardized tests, grades on high-school math courses, numbers of semesters of various kinds of high-school math courses). Then my colleague, Henry Walker, turned placement into a research project and wrote a system that automatically places students and even generates the placement letters. A decade or so passed, and a second research project transformed the original system into something a bit more usable.

It made sense that Henry did the placement when we were a joint department, particularly since he came to Grinnell as a faculty member in Mathematics. He continued to run the placement system after the two departments split because it served as a useful venue for advertising CS—students read their Math/Stats placement letters; it’s not clear that they would read CS placement letters, particularly since so few take CS in high school. Do we still need to advertise CS now that we are one of the most popular majors on campus? Believe it or not, but we’ve still found it valuable to advertise. I know that one of my favorite members of the class of 2020 took CS primarily because the placement letter suggested it [1].

When Henry moved to Senior Faculty Status, I inherited the system. This year is my third running it more-or-less on my own. I tell myself that it will take a bit of time. But things went so smoothly when I did the initial placements that I thought Oh, I’ve finally got things down well enough that it won’t be a lot of work. I was wrong.

After doing the initial placements, I found myself doing the following.

I met with the chair of Mathematics and Statistics to review students who had received transfer credit. We spent about an hour going through the placements associated with these students and thinking more generally about how we place transfer students and students with strong AP scores. That also contributed to further discussions of whether the placement system was still making the decisions that we thought it should make. We discussed one possible new decision and decided not to try to implement it this year. I’ll need to revisit the issue with their department some time through this year.

I spoke with the chair of Mathematics and Statistics about other issues in the letters. For example, we considered some language used in the letters and we identified which faculty members would be able to talk to students about their placement [2].

I then realized that there were some flaws in the expert system with regards to CS. In particular, I discovered that the system was encouraging students with weaker math backgrounds to start in our non-majors course even when they had multiple semesters of computers science. I thought that that recommendation would feel insulting to students. It’s also complicated by our inability to offer the non-majors course in 2017–18. I discussed appropriate placement with the department. I considered trying to update the rules in the system. Instead, I ended up manually re-placing [3] the students in this category.

I realized that the language about both CSC 105 and CSC 151 was incorrect. The former was incorrect in that we are not offering the course in 2017–18. It appears that I hadn’t been looking at that issue well last year since last year’s letter said, We will not offer CSC 105 in 2015–16 [4]. In the case of CSC 151, we are moving from a model of the course in which students ground their learning of computer science principles in problems in image making and manipulation to one in which students ground their learning of computer science principles through activities related to data science. That meant that I had to go through and update any of the descriptions of CSC 151. Unfortunately, we did not follow the DRY principle [5] in the code for generating letters. There may be a good reason. After all, we could be trying to convey subtly different ideas to different groups of students. But it meant that I had a lot to update.

I knew in advance that I would want to update the images in the letter. Why are there images? Well, the letter is only one-and-one-half pages long, which leaves room for something else. As part of the Advertise CS initiative as well as a broader Advertise the cool opportunities in Mathematics, Statistics, and Computer Science initiative, we’ve started including images from both CSC 151 and from student MAP projects in that space. It makes sense to continue that practice, even after the form of CSC 151 projects changes this semester.

That meant I had to dig out some images from my classes and from others, think about representation issues [6], remember how the system was set up to include images, figure out whose work the images represented, change the layout [7], and so on and so forth.

I also fixed some long-standing formatting issues in the letter, updated the list of people available for placement advising, dealt with language associated with the unavailability of some folks because of the eclipse [8,9],

When I showed a draft letter to my colleagues, they said something incredibly helpful, approximately

The kerning looks good for the regular face, but not so good for the bold face. You should fix that.

One of them did, fortunately, provide code that is supposed to make a difference [11]. I don’t see the difference, but I’ll hope that it’s there. I could, of course, compare a letter done without the extra kerning code to one done with it, but that’s not worth my time right now.

Henry and I have very different perspectives about the user interface for tools like the placement system. Henry likes to have programs print prompts, check the input, and then re-prompt the user if the input is incorrect. I would much prefer to provide the input on the command line and get an error message if the input is incorrect. Since I have to run the system for the next few years, I thought I should make at least one update this year, for the script that generates letters.

Here’s how it looks in the Henry version.

$ php generateLetters.php
Which year would you like to generate letters for?
2017
PHP Deprecated: mysql_pconnect(): The mysql extension is deprecated and will be removed in the future: use mysqli or PDO instead in /path/to/placement/system/mysql-connection.inc on line 7
Please enter sorting criteria: 
1 - name
2 - advisor, student name
3 - pobox
4 - studentid
1
[543 lines elided for confidentiality]
Total Number of students: 453

Here’s how it looks now that I’m done hacking at it.

$ php generateLetters.php 2017 1
PHP Deprecated: mysql_pconnect(): The mysql extension is deprecated and will be removed in the future: use mysqli or PDO instead in /path/to/placement/system/mysql-connection.inc on line 7
[543 lines elided for confidentiality]
Total Number of students: 453

There is, of course, more work to do. I should either take the filename for the letters as a command-line parameter or send the data to stdout. I need to fix the deprecated mysql_pconnect. But since I don’t use PHP except when working with the placement system, I don’t feel particularly encouraged to do so now.

I formatted the 453 letters [12], printed them [14], and brought them to student affairs for them to place in the students’ advising folders.

Of course, those aren’t the only letters to print. I’m also supposed to print a separate set to put in student boxes. You may have noted that the prompts above suggest that we can order them by box number. For some reason, that didn’t work with my new command-line hack. But I didn’t realize it until after I’d printed them, which meant that I started going through them to put them in order [16]. I discovered a variety of interesting issues. Two students have the same box number [17]. Students with apostrophes in their names lost the apostrophes. And maybe one more.

What happened with the apostrophes? I’m pretty sure that it’s something with how the PHP script processes the input file. I wrote to Henry and he sensibly suggested that it’s not a good idea to update a system during production time. I did, however, find the line that does it.

  //remove ' from field values
  $string = str_replace("'","",$string);

So that seems pretty damn intentional. I wonder why they made that decision in designing the system. But, like Henry, I don’t want to update a system while it’s being used. In any case, I went back and identified which students had apostrophes in their name and updated the database manually. I printed replacement letters, replaced them in the organized by box number pile, and brought a second set of replacement letters to student affairs to update the incorrect letters I had not noticed previously.

I asked the Registrar about a strange standardized test score and a related course equivalency. They told me to update the course equivalency. This may be the first time I’ve had to do so. It’s good that I know how to work directly in MySQL.

I then generated summaries [18] and placement sheets [19] and placed them in faculty boxes.

Am I done?

No.

I have a long-term commitment to putting everything on GitHub. That will allow [21] me to clean up the stuff a bit [22]. It will also make it easier for me to back out of some of the changes I’ve tried to make. Perhaps most importantly, it will let me more easily run the scripts on a virtual machine that has no other purpose than to serve as our placement system. That switch will help better protect student information.

But putting it on GitHub probably has to wait until a lull in the semester, particularly because I want to consider how to separate the confidential information from the code [23]. Maybe fall break.

I also need to get the final placement data to the Registrar’s office. I’ve given placement sheets to my colleagues. I’ll ask for those sheets back in a few weeks, enter any changes into the database, and send the revised placement data to the Registrar’s office [24].

Once I’ve done those two things, I’ll be done for this year.

Unless I decide to go back and fix the multiple issues I’ve just mentioned [25].

Maybe next year I’ll just let the Historian do it [30].

---

[1] That’s not the only way we get students who might not otherwise take CS. One of my favorite majors in the class of 2019 took CS only because their advisor cajoled them into it.

[2] Believe it or not, although we rely on the expert system for preliminary placement, we also want incoming students to speak with us to help make sure that the placement is as good as it can be. That’s become even more important in recent years as the College has shortened the deadline for students to add, drop, and switch levels.

[3] Placing again, not replacing.

[4] We did offer CSC 105 in 2016-17 and we hope to offer it again in 2018-19, although that version may be taught by an affiliated faculty member from another department.

[5] Don’t Repeat Yourself.

[6] That is, what does including particular images say about who does computer science, mathematics, or statistics and who participates in projects.

[7] We had relied on square images in the past. This year, I used two landscape-orientation images from summer MAP projects.

[8] No, we don’t have faculty members who are so terrified of the eclipse that they are cowering in their basements. We do, however, have faculty members who consider it important to go to a spot where they can view the 100% eclipse rather than 95% eclipse.

[9] The letter now indicates something on the order of If you want to speak to someone in CS, see X on Monday or Y on Sunday or Tuesday [10].

[10] Three of us are out of town for the Eclipse. Two of us are teaching Tutorial and should be spending Monday and Tuesday with their tutees. One of us is on leave and should be avoiding administrative work. One of us is brand new to Grinnell and should not advise students on placement. The numbers of available faculty don’t look good.

[11] Here’s what they sent me.

\usepackage{microtype}

%%% Set up microtype %%%
\microtypesetup{
  final,
  tracking=true,
  kerning=true,
  spacing=true,
  factor=1100,
  stretch=20,
  shrink=20
}

%%% No tracking for smallcaps %%%
\SetTracking{encoding={*}, shape=sc}{0}

[12] pdflatex letters2017.tex; it’s not hard and it’s relatively fast. Switching to pdflatex was one of the first things I did when taking over the system.

[14] Printing is less easy. Our printer freezes if I try to print all 906 pages [15]. So I have to break them up into groups of fifty or so pages. And then I have to keep track of which ones I printed.

[15] 453 double-sided letters.

[16] Yes, I eventually decided to ask our ASA to do that task.

[17] I reported that issue.

[18] We placed N students in CSC 151, M in CSC 161, etc.

[19] Stu Dent has recommended placement of 151 for CS, 208 [20] for Statistics, and 131 for Math.

[20] 208 is code for Take MAT 131 (Calculus I) and then take MAT 209 (Applied Statistics). We contrast it with 209, which is for students who have already taken Calculus I.

[21] encourage?

[22] There are way too many files that Henry or I created along the way.

[23] The confidential information includes not just the FERPA-protected information, but also the password and account for the MySQL database.

[24] If our ITS folk have their way, I’ll place the data on an encrypted flash drive and hand-carry it over. That actually seems like a reasonable approach to me.

[25] Have you kept count? I haven’t. Let’s see … (a) We’re relying on a deprecated SQL library. (b) My command-line hack for the generate letters script doesn’t seem to work for organizing the letters by box number. (c) The generate letters script should take the file name as a command-line parameter. (d) We need to figure out why they are intentionally removing apostrophes from the input. If there’s a good reason, I should write a more sensible script to fix the names the remove apostrophes code munges. (e) I should find more sensible ways to notate the more complicated placements [26]. (f) I should make more scripts work from the command line. (g) I should check on the kerning. Oh, so much fun [27]!

[26] Did I mention that one? One of my colleagues complained about the strange numbers. My understanding of the system suggests that it requires the numbers for some of the cascading decisions (e.g., If the math placement is about 131 …).

[27] I’m both serious and facetious. Since I enjoy writing code, there are parts of updating the code that I will probably enjoy. I even enjoy debugging because I find a sense of accomplishment when I figure out the cause of a bug [28]. On the other hand, since it’s someone else’s code written in a language I don’t regularly use, there are other parts I will most certainly not enjoy.

[28] The word I appears a lot in that sentence. But I like that sentence more than I even enjoy debugging because there is a sense of accomplishment that accompanies the discovery of a bug. Perhaps it’s because the Little Red Schoolhouse emphasized the value that readers find in active verbs rather than nominalizations [29].

[29] Yes, there are better ways to write that sentence, too. But it’s late.

[30] Feel free to ignore this inside joke [31].

[31] I’m relatively sure that a week from now, I won’t be one of the insiders [32].

[32] That means that I’ll have forgotten what the joke is about.

---

Version 1.1 of 2017-08-12.