What to Expect#
The Fish-PACE Hackweek will focus on applied, hands-on learning, with participants engaging in extended periods of small-group work. Our tutorials are designed to offer a broad snapshot of data science tools to support your applied investigations. Due to the relatively short duration of our events, we are not able to provide comprehensive, in-depth training in fundamental tools. Rather, our goal is to inform you about the types of tools we think are best suited to working with PACE datasets, with a focus on the Pangeo ecosystem of Python tools for big data geoscience. The details of implementation will be what you work out via peer-learning (helping each other) in your project group.
Typical Workflows and Tools#
Here are a few specific scenarios of how hackweek participants will engage with data science tools:
Connecting to a Jupyter Notebook environment and accessing content for tutorial training.
Accessing cloud-hosted remote sensing data using earthaccess and plotting it using matplotlib.
Exploring multi-dimensional remote sensing data using xarray.
Opening CSV tabular data in Pandas and run tools to conduct satellite matchups.
Modifying code, committing it to Git and pushing changes to GitHub, for others on your team to view and edit.
Exploring methods for high performance computing such as using Dask and parallelization
Preparing datasets for machine learning tools, including PyTorch and TensorFlow for neural networks
These are examples of the types of activities we will do at the Fish-PACE hackweek in a collaborative setting. Be aware that most of the project work will be within self-organized project teams. Much of the hackweek will be spent running code (via notebooks), writing code and talking about code. The mentors and organizers will provide links to tutorials and help trouble-shoot code, but much of the learning comes from working on a project together.
All tutorials will be in Python using the Pangeo ecosystem of tools for computing in the earth sciences. For participants wishing to brush up on their skills before the event, we recommend viewing the resources as described on the Pythia Foundations website. Teams are welcome to do their project in R and our compute platform fully supports R for earth science computation. The HackWeek mentors/helpers are experienced in Python, R and Matlab.
HackWeek Projects#
A good hackweek project is a concrete idea that a team can flesh out in a week together. Not everyone needs to code. There is background research to do, data to find, and lots and lots of data wangling. A big part of the fun of hackweek is working together with a group with a diverse set of interests and skills. “I’ll find some data.” “I make some maps of our study area.” “I’ll figure out how to do a boosted regression tree.” “I’ll use that tutorial we were shown and get xyz PACE data for our region.” etc, etc. It is messy, but through this process you’ll learn new skills and also get to know your project team mates.
The project work is a combination of
fleshing out a science idea that is small enough on Monday brainstorming.
dividing up into tasks so that everyone can participate.
coding and data wrangling on Tuesday through Thursday.
and then Friday, frantically putting a presentation on your project and results.
Checkout projects from other hackweeks to get an idea of projects done in earth science hackweeks