ORIE 3120 Spring 2022

Course Description

The practical use of software tools and mathematical methods from operations research, machine learning, statistics and data science. We begin by learning to manipulate data using structured query language (SQL) and visualize it using Tableau and Python. Then, continuing in Python, we learn the use of machine learning and statistical methods include multiple linear regression, classification, logistic regression, clustering, and time-series forecasting. We also learn about causality, how making decisions based on observational data can be dangerous and how to design experiments to avoid these dangers. Finally we learn about stochastic modeling and simulation using what we've learned from all of this data, and how it can be used to make decisions. These topics will be presented in the context of business and public sector applications in healthcare, transportation, manufacturing, retail, and e-commerce.

Prerequisite: ENGRD 2700

ENGRD 2700 may also be taken concurrently with ORIE 3120 as a corequisite. However, you will be responsible for material from ENGRD 2700, even if that material is taught in ENGRD 2700 after it is used in ORIE 3120.

Communication

The course website is http://people.orie.cornell.edu/pfrazier/orie3120/.

We use Piazza for announcements. You are responsible for reading all announcements and pinned posts made on Piazza.

Your primary point of contact in the course are the TAs for your assigned recitation. Please speak with them about any issues involving technical support, grading, or logistics. As much as possible, please channel questions through either live conversations (in person or over zoom) with TAs and Prof. Frazier, or via public Piazza posts.

If a question cannot be asked live or via a public Piazza post, then make a private Piazza post. If you feel that the issue should not be shared with the full TA team, make the private post to the head TA (Mark Gluzman) or to a TA from your recitation. This ensures centralized communication, which is critical for ensuring a timely response.

Please avoid sending email. Prof. Frazier receives a lot of email and much of it unfortunately falls between the cracks. To help the teaching team track and respond to communication about ORIE 3120, Piazza is much better. You'll get a faster reply if you make the post visible to more people. As much as possible please make private posts visible to the head TA (Mark Gluzman). It also helps us respond more quickly if it is visible to the entire TA team.

Can I get a permission code to add the course?

ORIE 3120 is a popular course, but we need to keep it reasonably sized to ensure that those taking it can be supported by the teaching team and get questions answered in a timely fashion. We are giving preference to ORIE and ISST majors, since they are required to take the course. Others can join the waitlist by emailing Heidi Russel (hjr27@cornell.edu) or joining piazza and making a private post to the teaching team.

Reading

Readings for the course are posted on the course website. There is no textbook.

Lecture Questions

Go to PollEv.com/orie3120 to answer questions for each day's lecture. During most of the semester, these must be answered during class. During the virtual instruction period, if you are in a time zone where lecture overlaps with hours between 10:30pm and 8am, make a private post on piazza within 24 of the lecture with your location and answers to the poll to get credit. Some lectures won't have questions.

Try to answer questions correctly. If you get at least 75% of questions correct during the semester, you will receive full credit for all days when you answered at least one question. If you get less than 75% correct, points will taken off.

You may choose not to answer up to two lectures' worth of questions without penalty.

In addition, all students will be given full credit for the lecture questions in the first two days of lecture.

Answering poll questions for someone else, especially if they cannot answer the questions themselves and ask you to do it for them, is a violation of Academic Integrity.

Recitation

The primary way to complete a recitation is to attend your recitation section, work on it collaboratively with your classmates and TAs, and individually submit your own work. To receive credit, submit answers to all recitation questions on Gradescope by the end of the last recitation in the week, and make sure that these answers are good enough to make clear that you made an honest effort to complete them well. It is ok to miss a question now and then, but try to answer them correctly.

You may do the recitation on your own instead of coming to your assigned recitation. To receive credit, you must submit answers to the questions on the recitation writeup through Gradescope by the end of the last recitation in the week. These answers should meet the same standard met by students attending the recitation. Of course, if you are having trouble completing the questions on time, you are still welcome to come to recitation and get help.

We suggest that you bring a laptop with you to your recitation section, so that you can work with your classmates and the TAs on lab assignments during that time. You can also use one of the lab computers if you prefer.

If your laptop ever malfunctions, note that there are many computer labs on campus, and the Cornell library even provides laptops that you can check out for short periods of time.

You may skip one recitation without grade penalty.

Some additional details for those that like fine print:

  • If you wish, you can verify with your TA or classmates that your answers are correct. This, however, is not necessary to receive credit. Full credit is given for solutions that represent an honest effort to solve all the questions well, even if not everything is perfect.
  • If you are having trouble with the recitation and fear that you won't be able to finish it by the end of your recitation, and are in attendance (in person or on zoom), then talk to a recitation TA. That TA will decide whether to award credit for the unfinished recitation. This judgement will be based on their assessment of whether you made an honest effort to complete the recitation. If you put in a good effort and are simply having difficulty, then the TA will award you full credit.
  • If you join the zoom or in-person recitation substantially later than its start, then this will be considered if the TA needs to make an assessment about the level of effort made. In general — don't come late to recitation.

Homework

Homeworks assignments should be completed by yourself, outside of recitation, and submitted through Gradescope. You may discuss the homework problems with other students, but your final submission must be your own work. We currently plan about eleven homeworks, although this may change. Your lowest homework score will be dropped.

Homework will be due weekly starting the week of Jan 31.

We recognize that students are balancing many priorities, and so we make two accommodations to allow for late or missed homework.

  1. Students have five slip days that may be used through the semester with no grade penalty. For example, you could submit one homework three days late and another two days late; or your friend could submit five homeworks one day late each. To calculate slip days, we round up: a homework submitted 24 hours and one minute after it is due will use two slip days. Homework submitted after all your slip days are used will receive a score of 0.

  2. We will discard your lowest homework score from the semester.

Extensions will be granted only if your college advisor issues a "request for academic consideration" (generally for medical reasons), or in other such exceptional circumstances. Requests due to job interviews, other classes and assignments, and poor planning will not be considered. We suggest you save your slip days and plan to complete all homework assignments to insure against catastrophe.

You can ask for your homework to be regraded up until four days after grades have been posted. Regrades can increase or decrease your grade. Regrade requests must be submitted through Gradescope. (We will regrade the whole assignment, not just a single question.)

Although some assignments are more difficult than others, we weight all assignments equally when dropping your lowest homework grade and when computing your overall homework score.

Homework and Slip Days: In general, we will apply slip days and choose the homework to drop that will maximize for grade. For most students, the number of late days is small enough that all homework can be counted once slip days are applied. For these students, dropping the homework with the lowest grade will give them the best grade, and this is how grades will be computed. For other students, it may be better to drop a late homework with a higher grade to save slip days. This will be worked out for students to maximize homework points earned.

Participation

Up to 10 points will be assigned to students for participation as follows:

  • Up to 5 points will be given for completing your recitation assignment (either during your assigned recitation or beforehand), as recorded by your TA.
  • Up to 4 points will be given for completing your lecture questions.
  • 1 point will be given for filling out the course evaluation at the end of the semester.
  • Students can make up lost points on class participation by answering questions from others on piazza.

Late policy and extensions

The five slip days and policy of dropping your lowest homework score, one recitation, and two days of lecture questions constitute our extension policy. We will make no exceptions to the policies stated above: late homework (beyond the five slip days) or late recitation/lecture question answers will not be accepted.

Here are a few words of caution about late submissions:

  • The deadline for an assignment is not the time by which you must finish writing a solution; rather, the deadline is the time by which you must successfully submit your solution file on Gradescope. We recommend that you submit your file at least one hour before the deadline.

  • You must submit your recitation and homework through Gradescope; email submissions, whether late or on time, will be deleted without response.

  • Gradescope enables you to upload as many times as you wish before submissions are closed. Only the most recent version will be graded. Requests to have the course staff grade earlier versions (with or without a penalty) will be denied.

  • It is your responsibility to verify before the deadline that you have submitted the intended versions of your files. Requests to substitute another version (e.g., "I accidentally submitted the wrong files") will be denied.

Project

The class will culminate in a final project in which you use the data visualization and data analysis skills you\’ve learned to analyze an interesting dataset. This project will give students experience in the kind of work that a data scientist might perform in industry, or at a government or nonprofit agency. These projects will be completed in groups of 2–3 students, with teams of 4 requiring a clear plan for dividing work and permission from the instructor.

There will be two major project milestones: Data visualization, tentatively due in the week of April 19; and Data analysis + stochastic modeling, due during the final exam period.

In these project milestones, you will be required to use data and tools from the class to explore and answer questions that you will identify. For each milestone, you will provide written reports that should clearly explain to others what you discovered, how you discovered it, and why it is important.

Both reports will be graded using a combination of peer-feedback and TA review. Write a report you think your peers will enjoy reading!

Exam

There will be only one exam during the course. This exam is scheduled for February 22.

This exam will cover material from lectures, recitations, and homeworks covered in the course by the point when the exam is given.

In the case of exam conflicts (see Cornell exam policy), or if you need extra time due to a disability, please notify the head TA at least one week before the exam.

You may ask for your exam to be regraded up until four days after grades have been assigned. Regrades may increase or decrease your grade.

If you cannot attend an exam because of a health or family crisis, or similar life events, and this occurs after the one-week cutoff to alert the head TA of exam conflicts, then you must still alert the head TA before the scheduled start of the exam. If existing makeup exam time(s) do not work for you, then the accommodation offered may be an individual oral exam. You must alert us in advance: don't just skip exams and assume you can be excused retroactively. If you miss an exam and we didn't excuse you, you get a zero for the missing score.

Academic Integrity

Absolute integrity is expected of every Cornell student in all academic undertakings. If you are unsure about what is permissible and what is not, please ask.

We encourage you to discuss your work with your friends and classmates. You will definitely learn more in this class if you work with others than if you do not. Ask questions, answer questions, and share ideas liberally.

Cooperation has limits, however, as set forth in these university and course policies:

Integrity includes you being honest about the sources of the work you submit. When you submit work in this course, you are representing it as the work of the stated authors, subject to any exceptions that are clearly stated in the submission itself. To avoid committing plagiarism, simply be sure always to accurately credit your sources. To do otherwise is to commit fraud by claiming credit for the ideas and efforts of others, and that is punishable under the Code of Academic Integrity. Penalties for violation of Academic Integrity may be severe, ranging from a zero grade for an assignment or exam, up to expulsion from the University for a second offense.

Grades, on the other hand, are about the course staff assessing what you have learned. If you turn in someone else's work for course credit, and forthrightly acknowledge you are doing so, you are not acting dishonestly and are not violating academic integrity. But you also give us no basis for concluding that you have learned the course content. So you are likely to receive a grade penalty, but it will be less severe than if you had failed to cite the other person.

We recommend the following rule of thumb: Never look at any other student's solutions (including source code), or have their solutions in your possession in any portion or form whatsoever. Once you have seen another solution, it becomes impossible to unsee and is likely to infect your own. Likewise, never share your solutions with other students. That includes not writing code together at a whiteboard: even if you erase it and later write code separately at a computer, you are likely to write similar code that could be flagged as a potential violation.

You are always free to use code presented in this class in lecture, or on this semester's course website. It does not require citation. Any other code, however, at least requires citation, and it could result in a grade deduction.

Grading

We expect the breakdown for the final course grade to be as follows:

  • Prelim 1: 25%
  • Homework: 30%
  • Participation (recitation + course evaluation): 10%
  • Project: 35% (equally split across the first milestone and the final project miletone)

Recitations, homeworks, and exams are weighted equally within their category unless otherwise specified. These weights are approximate. We reserve the right to change them later.

Students often want to predict their final letter grade. We will not publish details about how scores translate to letter grades. However, we can give a few rules of thumb. Suppose the numerical grades can range from 0 to 100 points. Generally scores in the 90s correspond to grades in the A range, scores in the 80s correspond to grades in the B range, scores in the 70s correspond to grades in the C range, etc. Historically the median grade in this class is around a B+.

Students taking the course with the S/U option should note that a grade of S (passing the class) correspond to a letter grade of C- or above. Generally, to obtain a passing grade, students take the exam, complete the project, and turn in all but two or three recitations and homeworks.

Sometimes students ask whether the final grade is curved. The answer is that it depends on what you mean by "curved." Any mapping from numeric scores to a letter grade implicitly defines some kind of curve. But we will not give out a fixed percentage of A's, B's, etc. In fact, we would be delighted to give a high grade to all students who complete all assignments and show mastery of the material on the exams.

We do not offer extra credit, nor will we consider any request to alter the grading scheme based on personal considerations.