OR 6205 · Deterministic Operations Research

Course Overview

Course: OR 6205 — Deterministic Operations Research (Graduate)
Role: Instructor of Record
Institution: Northeastern University, Dept. of Mechanical & Industrial Engineering
Term: Spring 2026 · 6 registered, 5 attending
Preparation: Teaching Fellowship, Fall 2025 — shadowed Prof. Himlona Palikhe in this course before teaching it
Award: 🏆 2026 COE Outstanding Graduate Teaching Award

This course introduces graduate students to the theory and application of deterministic operations research methods, including linear programming, duality theory, sensitivity analysis, network models, and integer programming. Students use Python (Gurobi) to solve real-world optimization problems and engage with the ethical dimensions of modeling.

Learning Objectives

By the end of this course, students are able to:

• Formulate linear and integer programming models from written problem descriptions
• Solve LP problems graphically, by hand (simplex), and computationally (Gurobi/Python)
• Analyze models using duality theory and sensitivity analysis
• Interpret optimization results in context and provide actionable recommendations
• Evaluate the ethical implications of modeling decisions — whose interests are represented, what is optimized, and what trade-offs are embedded in the model

Course Design Philosophy

This course was designed using backward design — learning objectives drove every assessment and in-class activity. Key design choices:

Active learning over lecture. Based on student feedback mid-semester, I transitioned from slide-based delivery to collaborative board work, step-by-step examples, and multi-format supplementary materials. Students said: “That’s so cool” — a moment that confirmed the shift was right.

Transparent assessment with rework. All assignments come with detailed rubrics. Students may revise graded work for additional credit, reinforcing that learning happens through iteration, not single attempts.

Ethics integrated throughout. Each case analysis asks students to consider whose interests are represented in the model and what unintended consequences might arise — grounded in Freire’s concept of critical consciousness.

Real-world case analysis. Each student identifies a real-world scenario that could benefit from LP optimization. Selected cases are worked through collaboratively in class.

Selected Student Feedback (TRACE Evaluations, Spring 2026)

“I did like that she was very happy to teach this course cause I could tell that she loves this topic so much, and I like that she was able to listen to our feedback to make the course better. Definitely, the group project was the most fun project to do.”

“The instructor was very well-prepared and explained concepts clearly. Class time was used effectively, and the feedback on assignments was helpful. They were also approachable and created a respectful and engaging learning environment.”

“The instructor did a great job creating an inclusive and respectful environment where everyone felt comfortable participating.”

TRACE ratings, every item. Six students registered, one never attended, and all five students who actually took the course responded. I publish the whole report rather than a selection from it, because a page that lists only the favourable half is not evidence of anything.

Reading the term: three accounts, kept separately

I teach the way I do research, so I will not collapse this into one tidy story. There are three sources of evidence about what happened in OR 6205 — the numeric report, what students wrote in their own words, and what I observed while running the course — and they do not fully agree. I keep all three below, in the order I formed them, and I mark the places where they contradict each other rather than quietly retiring whichever one lost. A course is not an experiment with a clean identification strategy, and five responses is not a sample. Disagreement between sources is information, not an error to be resolved.

Account 1

Written before the departmental and university comparisons were available.

All the assessment happened in class, and it happened constantly. Students worked small assignments in the room every single session, and I gave feedback and corrections on the spot, in the same class. The only work assigned outside class was reading. The evaluations track that design almost exactly: in-class sessions rated 4.40, above the department, while out-of-class assignments rated 3.80, well below. Students were reporting the course I actually built.

This was not a question of effort. Students reported spending 3 to 10 hours per week on the course outside class, median 5 to 7, with two of the five at 8 to 10. They were tested daily and they were putting in the time.

So why did challenge and perceived learning come in low? Because none of that daily work was graded. It was purely formative — can you solve this, where did it go wrong — with correction arriving within minutes, from me or from a classmate. I made that choice deliberately: I wanted students attempting problems without the fear of a mark attached. What I did not anticipate is that I had also removed the instrument students use to answer the question how am I doing? Correction in the moment answers “was this answer right?” It never accumulates into an answer to “am I getting better at this?” One student put the consequence plainly: too much theory relative to practice left them “less confident in my knowledge.”

The class was also small and unevenly prepared. OR 6205 is introductory by design, and this cohort of five included two doctoral students who had already met the material. In a class that size, students well past the intended baseline move an average on a question about challenge, and balancing what to prioritise was genuinely hard.

Account 2

One student described a mechanism I had not named. In class we would be working on one topic and then shift “very quickly to a different topic that becomes too high-level for me to understand.” Read against the numbers, that says I was moving between levels of abstraction without marking the transitions: the sessions themselves were engaging, and rated so at 4.40, but the thread running through them was visible to me and not to the room.

The same student named a second thing. The only work I set outside class was reading from a textbook they found confusing, and they asked for videos instead. Out-of-class assignments rated 3.80 and course materials 4.20 — while the material students rated highest in the entire report, at 4.80 against a departmental 4.45, was the online scaffolding library of videos, readings, and worked media I had built to sit alongside the sessions. My students were telling me I had already made the better version and had not assigned it.

Account 3

When the complete report arrived with departmental and university means on all nineteen items, one thing I had written turned out to be wrong, and two remain unsettled.

Contradicted. I had written that every item below the departmental mean measured whether students could tell where they stood, and that none of them was about the teaching itself. The full report does not support that. Used class time effectively came in at 3.80 against a departmental 4.42 — the largest gap on any instructor question, with a course median of 3 against a departmental median of 5. Preparedness rated 4.20 against 4.58 and clarity 4.20 against 4.36. Those are questions about what I did in the room. The claim was a reasonable inference from the four instructor items I had at the time and it does not survive the eight. I have corrected it rather than deleted it, because the reasoning that produced it is the same reasoning I would want a student to show me.

Unsettled, and I hold both. My design explanation and the students’ transitions explanation are not the same claim, and the evidence does not choose between them. Ungraded formative work would depress perceived learning without touching class time effectiveness; unmarked transitions would depress both. The transitions account fits more of the report. But I was in the room every session and the report is five responses, four of which left the relevant comment blank, so I am not willing to discard what I observed on that basis.

Also unsettled: the cohort explanation. Two doctoral students in a class of five genuinely does move a mean on a question about challenge, and that is arithmetic, not an excuse. Against it, the students who rated challenge lowest asked for more material and more practice, which is not what being over-prepared sounds like. I keep the explanation and I keep the objection to it, and I decline to let one number settle a question about five people.

What I changed

The accounts disagree about weighting; they converge on what to do. Each change addresses something all three sources point at.

I sequence and announce transitions. The day’s thread is stated at the start of a session and closed at the end, and I say out loud when a discussion moves from formulation to theory and back, so a change in abstraction is an announced event rather than something students have to detect while also learning the material.

I assign the scaffolding materials instead of the textbook. The videos and worked readings my students rated at 4.80 become the required out-of-class work, and the textbook becomes a reference rather than the path through the course.

I add a small number of low-stakes graded checkpoints, and some problems attempted unsupported before the correction arrives. The ungraded in-class practice stays, because it is doing real work. What is added around it is evidence students can read for themselves.

I differentiate challenge. An introductory course with mixed preparation needs a core path plus extension work for students who arrive ahead, whichever account of the challenge score turns out to be right.

Course Materials

These are the actual materials from Spring 2026, not summaries of them.

Course structure

• 📄 Syllabus
• 🗺️ Course map

The syllabus was revised in three moments during the course. First, in the introduction class, then in two moments further on as they understood the dynamic of the class better. This approach was great so I could change the syllabus with their values in mind, but also, I was able to adjust as I felt it would fit that small class size better. The counterpoint is that it became unclear for the students the criteria, since it changed so frequently, which came up in one of their feedback.

Assessment

• 📋 Project worksheet
• 📝 Project report framework
• ✅ Project rubric

In class

• 🎞️ Slides — Introduction to Linear Programming
• 🎞️ Slides — The Simplex Method, Part I
• 🔧 Gurobi and Python guideline

Reflection

This course was my most significant teaching experience to date, and the most instructive — partly because it did not go uniformly well.

With full instructional autonomy, I redesigned the delivery mid-semester based on student feedback, moving from slides to collaborative board work. That change worked on its own terms. Students who had been struggling began completing assessments independently, and students who had been quiet started contributing to discussion.

But my overall teaching effectiveness came in below the departmental mean, and I think the two facts are connected. Students found the sessions helpful, at 4.40 against a departmental 4.34, and simultaneously judged that I used class time less effectively than the department, at 3.80 against 4.42. Those are not contradictory once you read the written comments: individual sessions were worth attending, and the movement between them, and between levels of abstraction inside them, was not legible. The syllabus rated 4.00 against 4.56 for the same underlying reason.

Changing the course while it was running made the course harder to read. Every adjustment I made in response to feedback moved something the students had already planned around: what was going to be assessed, how, and when. I was responsive at the cost of being predictable, and my students paid for that in clarity.

That is the tension I most need to work on, and I would rather name it precisely than generalise it away. Responsiveness and clear expectations are not opposites, but they conflict in practice unless the course is designed for change: a syllabus with defined places where adjustment can happen, assessment criteria that stay fixed even when the activities move, and explicit communication whenever something shifts, so a change registers as a decision rather than as drift. Adapting in real time is still what I believe teaching requires. Doing it without destabilising the students it is meant to serve is the harder skill, and it is the one I am deliberately building.

The 2026 COE Outstanding Graduate Teaching Award was received for this course. I hold both things at once: the award, and a first term as instructor of record that showed me exactly where I still have work to do.