jimscchen / mids

Berkeley MIDS Course Work

Berkeley MIDS Course Work

W200 Python for Data Science *

Work in progress

W201 Research Design and Application for Data and Analysis

The first course of the program is a gentle introduction to data science.
We discussed the meaning of data science and think of potential projects (imaginations run wild).
We also learned about interacting with other teams that may be unfamiliar with data science.
No technical/hard skills covered in the course.

W203 Statistics for Data Science

Intro to statistics, and some basic probability. We learned (or reviewed) basic stat techniques, such as t-test, linear regression, logistics regression, etc. The work is done in R.

W205 Storing and Retrieving Data

W207 Applied Machine Learning

This course covers the basics of machine learning, mostly through sci-kit learning.
We played with MNIST(digit recognition), newsgroup data set(text categorization) and mushroom data set(clustering).
We used Theano for neural networks. The work is (obviously) done through python.
For the final project, we worked on facial keypoint recognition using Convolutional Neural Net.

W209 Data Visualization and Communication

A survey of common visualization tools: Tableau, Gephi, d3.js, and light HTML/javascript.
For the final project, we built a website explaining the basics of options trading.
Unfortunately a weak course--asynch has a lot of business presentation 101 material and provides little value.
Fortunately the course is getting revamped soon.

W231 Behind the Data: Humans and Values *

Work in progress

W241 Experiment and Causality

The course covers experiment design, and how to measure effects and identify causality, concepts critical for A/B testing.
Our final project is an experiment examining how positive/negative reinforcement can affect self-confidence.
Amazon Mechanical Turk and Qualtrics are used for the online survey.
The analysis (as well as other assignments) is done in R.
The final report can be found here.

W251 Scaling Up! Really Big Data *

Work in progress

W261 Machine Learning at Scale

W266 Natural Language Processing

The course covers common NLP tasks and relevant tools (neural net/tensorflow, dynamic programming).
All work is done in python, using a mixture of google cloud and PC.
We implemented Kneser-Ney, LSTM-RNN language models, Forward-backward, Virtebi and CKY algorithms.
Our final project is an open information extraction and information retrieval system on Marvel movie transcripts.
The report can be found here.

W271 Statistical Methods for Discrete Response, Time Series and Panel Data*

Work in progress

W210 Synthetic Capstone

Putting the learning together in a data science project (done in python).
We developed a predictive model for ICU readmissions based on MIMICIII data set.
We also built a front end to display important information and trends to help the doctors visualize patient data.
The asynch content for the course touches on data projects and products, as well as some team dynamics.
Over all the asynch content is a bit fluffy and overlaps with parts of W201.

* completed through MIDS4Life after graduation