18.650

Table of contents

  1. Course Info
  2. Realistic Prerequisites
  3. Subject Matter
  4. Course Staff
  5. Lectures
  6. Problem Sets
  7. Exams
  8. Resources
  9. Grading
  10. Advice to Future Students

Course Info

Class Size 135
Hours/Week 8.1 (49 responses)
Instructors Tyler Maunu (Lecturer), Ashwin Narayan (Recitation Instructor), Jonathan Tidor (Recitation Instructor)
# of Responses to Course 18 Underground Questions 12/52 [CHANGE THIS]

Realistic Prerequisites

  • Consider 18.600 (or a substitute like 6.041) a hard prerequisite.
  • Students should be very comfortable with material from 18.02.
  • Some students found linear algebra experience from 18.06 and knowledge of convergence from 18.100 helpful.

Subject Matter

  • Mostly theoretical, but provides some real-world applications.
  • Fundamentals of statistics are covered thoroughly.
  • Students learn to “confidently set up a statistical model for a given problem and derive useful statistical tools like estimators, confidence intervals, and hypothesis testing.”
  • Some felt that the mock datasets felt a bit meaningless. Some also felt that results could have been proved more thoroughly.

Course Staff

  • Caring, engaging, and invested in the students’ learning.
  • Prof. Maunu is very understanding, and was very active in answering questions on Piazza.
  • Office hours, recitations, and review sessions were very good.
  • Jonathan was especially helpful, consistent, and patient when explaining concepts to students, even staying late after office hours.

Lectures

  • Lectures were engaging, but a bit slow.
  • Prof. Maunu works through guiding examples during the lecture, which students found helpful.
  • At times, there was not enough time in the lecture for Prof. Maunu to finish teaching the material or give enough guiding motivation for results.

Problem Sets

  • There was a problem set due every two weeks.
  • The problem sets were fairly challenging, but followed the lectures closely.
  • Students found them important for mastering the course material.
  • Problem sets were somewhat lengthy, taking some students up to 10 hours, but others about 6-8 hours.
  • Some students found later problem sets, which contained more calculations, tedious.

Exams

  • Difficulty was reasonable. Problems required little creativity and closely matched material covered in problem sets.
  • Exams were open book and open notes.
  • Due to the 12 hour time limit, exams were not stressful. However, they were designed to take 90 minutes to complete.
  • Some students found it easy to make silly computation mistakes and lose points for them.

Resources

  • There was no textbook. While the professor recommended an optional, supplemental textbook, students sorely missed having an official textbook.
  • The main resource was lecture slides annotated by Prof. Maunu, which students often found inadequate.

Grading

  • Very lenient and fair.
  • Students performed quite well on exams. Many grades were in the 80s and 90s.
  • Partial credit can sometimes be hard to predict, and cutoffs were not very transparent.

Advice to Future Students

  1. Take the probability prereqs for this class or study up beforehand, go to office hours, find some PSet buddies.
  2. Have a good understanding of probability.
  3. If you wanted to take this class because statistics is the foundation of machine learning, I think you’ll also need to take some more class beyond this one. This class is a reasonable intro that would probably help with more advanced classes though.