ORIE 5355/INFO 5370: Applied Data Science - Decision-making beyond Prediction

  • Instructor: Prof. Nikhil Garg (he/him), ngarg@cornell.edu
  • Semester: Fall 2026
  • Credits: 3 hours, Letter Grade
  • Lecture Timings: Mondays/Wednesdays, 2:55PM - 4:10PM [In person, Bloomberg 131]
  • Instructor office hours: Wednesdays, After class [Bloomberg 454]
  • TA office hours: TBD
  • Course frequency: twice weekly

Important links

Course Description

This course considers the data science challenges beyond training an accurate predictive model, especially for systems about people (data of behavior), and for people (deployed models to influence behavior). Whether for online marketplaces, transportation, governmental, or urban systems, effective data science in such settings requires dealing with user incentives and strategic behavior, networked and decentralized decision-making, and privacy and ethics concerns. This course is about all the ways introductory statistics/data science/machine learning fails when deployed in such systems, and how to nonetheless build effective systems.

Course objectives and learning outcomes

This course covers topics necessary to effectively analyze data and deploy decision-making models in such systems, providing both conceptual frameworks and hands-on data science practice. Core topics include every step of the data collection, analysis, experimentation, and model deployment pipeline. A running theme will be the practice of designing systems with privacy, fairness, ethics, and transparency in mind. We will motivate challenges and demonstrate solutions from across domains, including recommendations, matching, pricing, scheduling, and ratings and feedback systems. A guided data science project will connect concepts to practice.

This course is complementary to the other data science, machine learning, and operations courses at Cornell Tech. It is accessible to students who have not taken those courses yet, but also interesting and novel to those who have. Some topics (such as pricing, recommendations, and experimentation) might overlap with those covered in other courses, but the focus here is on practice and on the particular challenges of socio-technical systems – and the overlap is minimal. The course content and assessment is a mix of conceptual, theoretical, and programming-based.

Consult here for a guide to how this course relates to other data science courses at Cornell Tech.

Student outcomes: Students will be able to:

  • Articulate common challenges in data science about and for people, including incentives, strategic behavior, privacy, fairness, and ethics.
  • Analyze data collection, recommendation, pricing, experimentation, and deployment problems using appropriate conceptual and technical frameworks.
  • Design, develop, evaluate, and communicate data science models and systems that account for decision-making challenges.

About the instructor

Nikhil is an assistant professor of Operations Research and Information Engineering at Cornell Tech, whose research is at the intersection of computer science, economics, and operations – on the application of algorithms, data science, and mechanism design to the study of democracy, markets, and societal systems at large. Things he’s worked on include 311 reporting systems, surge pricing, rating systems, how to vote on budgets, gerrymandering, stereotypes in word embeddings, and political polarization on Twitter. Outside of academia, Nikhil has been a data scientist at Uber, collaborated with Upwork, other freelancing marketplaces, several government agencies, and led campaign data science at PredictWise during the 2020 US election cycle.

Prerequisites and Preparation summary

The homeworks will require use of Python, and the course assumes basic knowledge; in particular, from the beginning we assume that you are able to (or can pick up on your own) load and manipulate datasets in Python using Pandas and Numpy and calculate summary statistics such as weighted means. Please refer to the first homework assignment as a gauge for the level of Python we expect. We also assume some knowledge of statistics or data science: that you are familiar with introductory statistics concepts such as means, bias, variance, distributions, and linear algebra concepts such as matrix multiplication, vector norms, and dot products.

Corequisites

Applied Machine Learning or Urban Data, or equivalent, is required as either a prerequisite or a corequisite. In particular, we will assume after the initial part of the course that students have been taught the broad ML/data science modeling fundamentals that those courses teach, including, for example, how to implement various models in Python notebooks and to evaluate their performance.

Class and Laboratory Schedule:

Lectures: 2 75 minute lectures per week

Recitations: None required. Optional office hours with TAs and instructor

Textbook(s) and/or Other Required Materials:

None required to purchase. All readings will be distributed throughout the semester. All software used will also be open source/freely available.

Course communication

Course communication will primarily be over Ed Discussion, office hours, and email.

  • Ed Discussion: First resource for any question, whether regarding technical content or logistics. Please make your question visible to everyone, so that others may answer or benefit from your question. Instructors will aim to respond to questions in a 48-72 hour period, except for those of an urgent nature (e.g., typos on homeworks or lecture notes, clarifying course logistics, etc.). Among other things this means you should not wait until the last few days before an assignment is due to message us; we may not respond in time. You are encouraged to answer questions from other students, especially during the instructor “waiting period.”
  • Office hours: You are strongly encouraged to come to office hours for any reason. Office hours are the best way to ask in-depth technical questions, whether directly related to the course content or just things you’re interested in. TA office hours will often cover the homework questions in depth.
  • Email: Only for private questions and concerns, such as requests for accommodations. Please include “[ORIE5355]” in the subject line of any email. Technical questions will not be answered over email – please use Ed Discussion.

Please follow these norms in all communications with the instructors and other students.

Typical course topics covered with schedule

The meeting-by-meeting schedule is maintained on the course schedule page.

  • Data collection
    • Data constructs, surveys, ratings, polling, and implicit data exhausts
    • Challenges and biases: censoring, strategic reporting, social desirability, ratings inflation, privacy, etc
    • Technical solutions to challenges: stratification, weighting, post-processing
    • Non-technical solutions and case studies
  • Recommendations
    • Collaborative filtering and personalized recommendations; individual vs demographic based recommendations
    • Recommendations in practice: Capacity constraints, matching, 2-sided fairness, and other challenges (such as limited + missing data)
  • Algorithmic pricing
    • Basics of posted price mechanisms, algorithmic pricing
    • Personalized and dynamic pricing in practice (online marketplaces, supply/labor side wages, and roadway congestion pricing)
    • Fairness, ethics, and limitations
  • Experimentation
    • A/B testing basics
    • Experimentation in practice: networks, interference, clustering, experimentation over time, switchbacks, 2-sided experimentation, trade-offs across experiments
    • Ethics and communication of experiments
    • Introduction to causal inference without experiments
  • Miscellaneous: Exact topics based on student interest
    • Algorithmic explainability and transparency
    • Performance drift, strategic reactions to your model, Data feedback loops
    • Human-in-the-loop machine learning
    • Fairness audits and interventions
    • Differential privacy

Assignments, Exams and Projects

  • In class quizzes: 30%. Four in class quizzes. Lowest score dropped. Remaining quizzes are an equal part of the quiz grade. Quizzes cover class, homework, and other assignment content.
  • Homework: 30%. Three homeworks. Each HW is an equal part of the homework grade. Lowest score replaced by quiz average (after lowest is dropped).
  • Final project: 25%. Programming-based project coding an algorithmic agent that does recommendation and/or pricing. Submission involves 2 parts of code submissions, a written project report, and an oral exam.
  • Participation and Miscellaneous: 15%. Attend online/in-person class and have meaningful participation in the class community. Complete the occasional class survey, as well as the final official course evaluation.

Detailed HW assessment is available on the homeworks page, and project assessment is available on the project page. We note that, historically, the Final Project plays a substantial part of student grades, and homework grades tend to be high.

Tentative assignment deadlines (these dates are unlikely to be earlier, but may be later). Quizzes will either be in class or posted online on the stated date:

  • HW 1 – 9/15
  • Quiz 1 – 9/16
  • HW 2 – 9/29
  • Quiz 2 – 10/5
  • HW 3 – 10/20
  • Quiz 3 – 10/26
  • Project Part 1 – 11/12
  • Quiz 4 – 11/18
  • Project Part 2 code – 12/6 at 5pm
  • Project report – 12/12 at 5pm

Basis of grade determination

Letter grading. Grades will be at least as generous as the following, with A+ given at the instructor’s discretion.

  • A: 94-100
  • A-: 90-93
  • B+: 85-89
  • B: 80-84
  • C: 70-80
  • F: < 70

Assessment

Primary assessment is based on 3 homeworks, 4 quizzes, and a project. We will replace your lowest homework grade with your quiz average (if it is higher).

You will be allowed four total late days during the semester for homeworks. These late days allow you to turn in homework up to 24 hours late, with no penalty; you may also use all the late days on a single homework assignment (allowing you to turn it in up to 96 hours late), but that will leave you with no late days for other homeworks. When submitting an assignment late, please mark at the top how many late days are used for this assignment and how many you have used before, if any. Late days cannot be used for quizzes or the project – only homework.

Regrade, late assignments, and extensions policy

Research has shown that regrade requests lead to disparities due to differences in who is comfortable asking for regrades. Thus, regrade requests will only be accepted for clear, unambiguous errors in grading, such as when an assignment (or part of it) is marked missing when it was submitted on time in the manner instructed. Unambiguous errors on our part also include marking wrong an answer that is completely correct (code works, it’s the right thing to do, gets the right answer). However, it does not include judgement calls, such as whether a “right enough” code deserved 1 pt or 2 pts.

In the case of grading errors, however, we encourage you to come forward – please submit a regrade request on gradescope as soon as possible, within one week of receiving the grade. (For assignments/grades not submitted over gradescope, you may submit your request over email to the instructor and TAs).

For the same reason, we will not accept requests for extensions on assignments besides the late days policy above. We are hoping that between dropping the lowest scoring homework and quiz and the late days policy, you will have enough flexibility to accommodate your scheduling needs during the semester – without inducing disparities due to differences in who is comfortable asking for regrades and extensions.

Illness and attendance policy

The course will follow Cornell and Cornell Tech. Bottom line: don’t attend if you’re feeling ill, and we’ll figure it out from there. Otherwise, regular attendance is expected in class. Live remote attendance is not an option, and we will generally not be recording lectures and making them available.

80% of the participation and miscellaneous grade will be based on attendance, as judged by in-class sign-ins. Students who attend 80% of the days with sign-ins will receive full credit for that component, with the grade falling linearly from there. The remainder of the participation and miscellaneous grade will be based on filling out class surveys, completing the final course evaluation, and meaningful participation in class, on Ed Discussion, or in office hours.

Important Do not sign into class via the attendance links if you are not present in class. We will use the in class sign-ins to call on students to answer questions or otherwise contribute to class. If you are not present when you are called on, you may lose up to 5% from your overall course grade for each such instance, at the discretion of the teaching staff (up to the full participation + miscellaneous component of the course grade).

LLM usage policy

Our usage policy is as following:

AI-FREE icon: no AI for any use, unless otherwise specified Writing/conceptual work: AI-FREE. Do not use LLMs or similar tools for written conceptual answers or the project report.
UA icon: use with attribution Coding/programming work: UA. LLM use for coding help is allowed only with attribution, as described below.

Icons from the Cornell Center for Teaching Innovation AI Course Policy Icons.

  1. You are not allowed to use LLMs/similar tools for writing, such as to answer the conceptual/qualitative questions on homeworks or at all for the project report.
  2. You are allowed to use such tools for coding help. Your are discouraged from using it for the programming assignments, but may use it with attribution – the homeworks are meant to help you understand the ideas in lecture. For example, if you used ChatGPT to help you write a function, please add a comment like # Helped by LLMs above the function definition, with information on how it was used. If an agent/LLM completely answered a questions, you should include the comment # Mostly/fully written by LLMs This is to ensure transparency and academic integrity in your work. Attributing AI use is a part of each assignment, and an important skill: there may be a 1 point deduction for each problem part in which AI usage is not disclosed.

Furthermore, you must disclose use of LLMs when asking TAs for help: they will be given strict instructions to not help debugging code that has been generated by an LLM, and to instead ask you to debug it yourself. They will be instructed to ask you to explain the code and how it works and your intentions while writing it. If you are unable to do so, they will not be able to help you with the code.

LLM usage without careful thinking and feedback hinders your learning, as shown by substantial research. Many classes/faculty report low in class assignment grades due to students who over-rely on AI for their at home assignments.

You are ultimately responsible for understanding everything that you submit. For example, if you are unable to explain your code and strategy in the project oral exam, you will receive a poor grade. We further note that the written components of the assignments, such as conceptual questions on homeworks and the project report, will be graded in part based on creativity, originality, clarity, and conciseness – traits that are not strengths for LLMs. We anticipate LLM usage for such assignments to result in poor grades.

We also believe that responsible usage of LLMs can aid in learning. For example, you may use them to test your understanding of the course material, come up with and grade your practice problems, or to brainstorm and get feedback on your ideas.

We reserve the right to change this policy during the semester.

Academic Integrity

Each student in this course is expected to abide by the Cornell University Code of Academic Integrity. Any work submitted by a student in this course for academic credit will be the student’s own work. The policy can be found on the university’s website here: https://theuniversityfaculty.cornell.edu/dean/academic-integrity/code-of-academic-integrity/.

You are encouraged to study together and to discuss information and concepts covered in lecture and the sections with other students. You can give “consulting” help to or receive “consulting” help from such students for homeworks and the class project. Except in the case of group homeworks or projects, this permissible cooperation should never involve one student having possession of a copy of all or part of work done by someone else. For group assignments, no group should have in their possession work done by another group. On the top of each assignment, please list everyone with whom you discussed the assignment. Class quizzes must be completed wholly independently, with no help of any kind between students.

No electronics or talking will be permitted for the in-class assignments. You may bring and use printed notes.

Academic Freedom and Building Trust in the Classroom

Each person in this class is expected to respect the principles of academic freedom for instructors and classmates and will maintain the privacy of the classroom environment, as outlined in Cornell’s S20 Commitment to Academic Integrity, Equitable Instruction, Trust, and Respect, here: https://bpb-us-e1.wpmucdn.com/blogs.cornell.edu/dist/3/6798/files/2020/03/HonorFinal.pdf.

This commitment to building respect and trust in the classroom means members of this class will not: record, photograph, or share online any interactions that involve classmates or any member of the teaching team. Students will also respect the intellectual property rights of the instructor, and will not share or otherwise make accessible any course materials to anyone not enrolled in the course, without the instructor’s written permission.

This policy is not meant to restrict students’ ability to use classroom recordings in ways beneficial to their learning. Students who may benefit from recorded lectures and lecture playback, including students who use English as an additional language or who have accommodations from SDS, should speak to the course instructor to maintain transparency and trust in the classroom. Students approved to record lectures are expected to maintain the respect and privacy of the learning environment, as stated above. Students will also not enable anyone not enrolled in the course to participate in any activity that is associated with the course. Exceptions to this require the instructor’s written permission.

Academic Misconduct

A faculty member may impose a grade penalty for any misconduct in the classroom or examination room. Examples of academic misconduct include, but are not limited to, talking during an exam, bringing unauthorized materials into the exam room, disruptive behavior in the classroom, and unauthorized or undisclosed use of AI tools.

Should copying occur, both the student(s) who copied work from another student(s) and the student(s) who gave material to be copied will both automatically receive a zero for the assignment. Penalty for violation of this Code can also be extended to include failure of the course and University disciplinary action.

Students with Disabilities

Your access in this course is important to us. Please provide your Student Disability Services (SDS) accommodation letter early in the semester so that we have adequate time to arrange your approved academic accommodations. If you need immediate accommodations for equal access, please speak with us after class or send an email message to us and/or SDS at sds_cu@cornell.edu. If the need arises for additional accommodations during the semester, please contact SDS. You may also feel free to speak with the Student & Academic Affairs team at Cornell Tech who will connect you with the university SDS office. If you have, or think you may have a disability, please contact Student Disability Services for a confidential discussion: sds_cu@cornell.edu, 607-254-4545, or complete a disability self-disclosure form. You must request your SDS accommodation letter no later than 3 weeks prior to needing it.

  • Students currently registered with SDS: Once you request your accommodation letter and it is approved by SDS, it will be emailed to both you and your instructors. Processing time can be up to 48-hours.
  • Students not registered with SDS: The registration process for new accommodations can take up to three weeks. Once you are approved by SDS for accommodations, you will be able to request your accommodation letter for this course.
  • If you are approved for accommodations later in the semester: you must request your accommodation letter as soon as possible.

Religious Observances

Cornell University is committed to supporting students who wish to practice their religious beliefs. Students are advised to discuss religious absences with their instructors well in advance of the religious holiday so that arrangements for making up work can be resolved before the absence.Students are encouraged to anticipate their religious/spiritual needs early in the semester, and at least two weeks before the observance, leaving plenty of time for the professor and student to reach a reasonable accommodation.

It is your responsibility as a student to communicate your religious/spiritual needs to faculty and staff. Here are some important factors to guide you in that process:

  • Communicate early with your professors. For example, it is not reasonable to ask your professor to reschedule an exam just days before a major religious observance, especially those that are scheduled well in advance. Students are encouraged to anticipate their religious/spiritual needs early in the semester, and at least two weeks before the observance, leaving plenty of time for the professor and student to reach a reasonable accommodation.
  • Keep in mind that a reasonable accommodation may not be your preferred accommodation. Faculty must balance numerous factors in making accommodations, including the academic requirements of the course or program, their academic responsibilities outside the classroom, the availability of proctors, other important dates, etc.

Mental Health & Well-being

Your health and wellbeing are important to us, and you should always feel free to reach out to us for support. There are services and resources at Cornell designed specifically to bolster student mental health and well-being. Remember, your mental health and emotional well-being are just as important as your physical health. If you or a friend are struggling emotionally or feeling stressed, fatigued, or burned out, there are many campus resources available to you:

Cornell Tech students: This link provides a list of resources for Cornell Tech students: https://mentalhealth.cornell.edu/get-support/tech. You can additionally contact studentwellness@tech.cornell.edu with concerns.