Skip to content

Latest commit

 

History

History
123 lines (86 loc) · 9.09 KB

CPC-Spring2022-Syllabus.md

File metadata and controls

123 lines (86 loc) · 9.09 KB
Raw

Image of CU-stat

STAT GR5243/GU4243 | crosslisted in EESC

Climate Predication Challenges

A course offered by the Center for Learning Earth with AI and Physics (LEAP), an NSF Science and Technology Center (STC)


Course Information

Prerequisites

The pre-requisite for this course includes working knowledge of high school science and math, statistics and probability, data mining, statistical modeling and machine learning, and fundamental terminologies of earth systems and climate science. Prior programming experience in R or Python is required. Videos for crash courses in machine learning and programming, and introduction to climate science can be found here.

Description

This course is a project-based learning (PBL) course where teams of climate science and data science students collaborate to create machine learning predictive models for challenges inspired by LEAP's research. Students from different background will apply their prior knowledge, work together and teach each other in high-paced collaborative projects. Through a sequence of mini-projects, i.e., “challenges”, this course provides students a deeper understanding of using machine learning for climate science and support predictive capabilities. It provides training on a broad set of practical skills for climate data science research (e.g., handling geoscience data formats, data curation, cleaning and transformation, building ML workflow, and collaboration using Google drive, Google Colab, Git and/or GitHub). It will also offer discussions on the opportunities and challenges of using climate science and projections in decision processes.

No formal instruction on statistics, data science, machine learning, or climate science will be given. Project cycles run every 4 weeks, where we will have mini-group data projects. Groups will be formed randomly with students from both climate science and data science background. Project products will be peer-reviewed, in addition to evaluation by the instructional team.

Course organization

This course will have a total of three project cycles. Each project cycle follows a sequence of four types of activities.

a. Dataset/challenge release, introduction to climae data science problem, individual exercises, team forming

b. Lecture/tutorial

c. Brainstorming, live hacking, code sharing

d. Team presentation, peer reviews, within-team peer reviews

Students will be working in teams of 4-5 students that will be randomly formed. For a meaningful experience in climate data science, students are expected to collaborate and work together on all the stages of a project. Code sharing and brainstorming are great opportunities to learn from each other.

We will have a total of three project cycles for this course (topics are subject to change):

  1. [Group] Reproducible notebook for exploratory data analysis of climate data sets
  2. [Group] Physics-informed machine learning
  3. [Group] Predictive modeling with uncertainty quantification

Below is a tentative schedule we will follow.

  • Week 1 (Jan 18): 1a+1b
  • Week 2 (Jan 25): 1b+1c
  • Week 3 (Feb 1): 1b+1c
  • Week 4 (Feb 8): 1d+2a
  • Week 5 (Feb 15): 2b+2c
  • Week 6 (Feb 22): 2b+2c
  • Week 7 (Mar 1): 2b+2c
  • Week 8 (Mar 8): 2d
  • Spring break: no class
  • Week 9 (Mar 22): 3a+3b
  • Week 10 (Mar 29): 3b+3c
  • Week 11 (Apr 5): 3b+3c
  • Week 12 (Apr 12): 3b+3c
  • Week 13 (Apr 19): 3b+3c
  • Week 14 (Apr 26): 3d

Evaluation

Students' performance will be evaluated based on

  • [85%] Project products (instructor-reviewed and/or peer-reviewed, averaged over 3 projects). Each project description will have explicit grading rubrics.

  • [15%] Individual participation (based on individual tasks and instructors' observation).

    A note on participation evaluation.

    In addition to individual tasks such as peer reviews, for each project, we will enforce formal evaluation of participation as follows.

    • Each project needs to have clearly communicated collaboration and task assignments, as part of a contribution statement for each project submission.

    • Students should participate actively in class discussion and online discussion.

    • We will give participation score for each project cycle, the average of which will contribute to 15% of your final grade. The participation will be graded on the following curve.

      • A (1.8-2): project leader, major contributor who contribute substantially in every stage of the project and class discussions.
      • A- (1.5-1.8): major contributor who contributed substantially to two stages of the project and some discussions.
      • B+ (1.2-1.5): average participation, participate in the discussion at every stage and contribute substentially in at least one stage of the project and some discussions.
      • B (1-1.2) or lower: below average performance.

    • This is to ensure a positive learning process for all of us.

Communication

Projects grades are managed in courseworks. We will be using the courseworks' discussion/announcement tools for our class communication and discussion. The system is highly catered to getting you help fast and efficiently from classmates, the TA, and instructors. Rather than emailing questions to the teaching staff, we encourage you to post your questions online.

Textbook

There is no required text. As part of this course, we will learn from what we can find online and in academic papers. Here are a couple of recommended reference books.

Class policy

  • We learn together through projects. Please stay positive and congenial. Share what you know with your peers and also learn from them.

  • Working towards deadlines can be stressful. Remember, emails or online posts do NOT have tones. Be mindful about how you phrase your questions, comments, inquries and suggestions. Also be generous when reading them.

  • Academic Integrity is the cornerstone of meaningful teaching and learning. It is especially important for our project-based course. Remember what matters more is how much you learn not what grade you will get. In your project, document references and resources that have been incorporated into your project and accredit them appropriately. Plagiarism is one of the most likely forms of cheating in this course.

  • Be a good team member and contribute to each project as much as you can. Don't underestimate the efforts of your teammates. Something seems simple may not be that simple.

  • Emails related to learning and projects shall be redirected to our discussion board.

  • Students are expected to check emails at least once every 12 hours during the week and every 24 hours over the weekend. Students should make sure not to miss any important class-related announcements sent by emails or posted on Courseworks. Emails will be delivered to the students' official UNI. It is the students' responsibility to ensure that these emails are properly forwarded if they choose to use an alternative email address.