Python for Data Science (AY250)

Homepage

Welcome to Python Computing for Data Science
An Undergraduate/Graduate Seminar Course at UC Berkeley (AY 250)

CAMPBELL HALL: Friday 1 - 4 PM FALL 2016

Synopsis

Python Links to an external site. is becoming the de facto superglue language for modern scientific computing Links to an external site.. In this course we will learn Pythonic interactions with databases, imaging processing, advanced statistical and numerical packages, web frameworks, machine-learning, and parallelism. Each week will involve lectures and coding projects. In the final project, students will build a working codebase useful for their own research domain.

This class is for any student working in a quantitative discipline and with familiarity with Python. Those who completed the Python Bootcamp Links to an external site. or equivalent will be eligible

Course Schedule

Date Content Reading Leader
Aug 26 Advanced Python Language Concepts (decorators, OrderedDict,
Generators, Iterables, Context Managers)
- GIT Links to an external site.
- scipy §2.1 Links to an external site.
Josh
Sep 2 Pandas, Scipy, & Numpy - scipy §§ 1.3 Links to an external site., 1.5 Links to an external site.,2.2 Links to an external site.
- numpy Links to an external site.
Josh
Sep 9 Data vizualization (Matplotlib, Bokeh, Altair, Plotly, mayavi) TBD Josh
Sep 16 Interacting with the world (requests, email, IoT/pyserial) TBD Josh
Sep 23 Parallelism (asyncio, dask, IPython cluster) TBD Josh
Sep 30 Database interaction (sqlite, postgres, SQLAlchemy),
Large datasets (xarray, HDF5)
TBD Josh
Oct 7 Image processing (OpenCV, skimage) TBD Stéfan van der Walt
Oct 14 Machine Learning I (sklearn, NLP) TBD Josh
Oct 21 Machine Learning II (keras [tensorflow]) TBD Josh
Oct 28 Bayesian programming & Symbolic math TBD Brett
Nov 4 Web frameworks & RESTful APIs, Flask TBD Josh
Nov 11 holiday
Nov 18 Computational Frameworks (Docker, AWS, Azure, AWS-Lambda) TBD Josh
Nov 25 holiday
Dec 2 Speeding it up (Numba, Cython, wrapping legacy code) TBD Josh
Dec 5/Onward final project work
Onward

Sidebar Concepts

Throughout these lectures we will be peppering in sidebar knowledge concepts:

  • Jupyter & JuypterLab
  • using git & github
  • Docker
  • Data science workflows
  • reproducible research
  • application building
  • debugging
  • testing

Workflow

Each Friday we will be introducing a reasonably self-contained topic with two back-to-back lectures. In between a short (~20 minute) breakout coding session will be conducted. Homeworks will require you to write a large (several hundred line) codebase.

Email Josh with any questions if you cannot attend.

Contact

Email us at ucbpythonclass@gmail.com or contact the professor directly (jbloom@astro.berkeley.edu). Auditing is not permitted by the University but those wishing to sit in on a class or two should contact the professor before attending.