NYT data scientists open online courses, quickly get started with Processing and D3.js for visualization

Graduate students do not write papers at home on weekends, then take some online courses with high CP value

I like the life of staying at home on weekends and not going out to socialize the most. As a postgraduate student who is so homey but doesn't want to write a thesis, the self-study progress of online courses is very fast. It would be nice if I could give myself some energy to write a thesis...

REVIEW

• The New York Times' Data Analysis Methodology

• Data Analytics class at Columbia University by New York Times data scientist Chris Wiggins

• Processing animation series by Danish artist Tim Rodenbröker
• Learn D3.js data visualization, get a certificate by the way

NYT's chief data scientist deciphers NYT's analysis method

Christopher Wiggins is the leader of the data analysis team at The New York Times. I happened to see a new video released by Towards Data Science's YouTube a few days ago, and when I clicked on it, I found out that it was the CW's speech at the Toronto Machine Learning Summit. The film mainly shares the methodology of "The New York Times" applying machine learning. Although it is a methodology, CW's explanation is approachable. Even people who have no concept of machine learning can understand the function of data analysis in machine learning. Maybe this has something to do with CW itself teaching at Columbia University, offering data analysis classes for students who don't have any data foundation?

takeaway

1. The data visualization team of The New York Times is the face, and the data analysis team is the cornerstone. The data analysis team led by Chirs Wiggins is responsible for providing answers to business models such as business operations and subscription systems.
2. There are usually three ways of interpretation by the data analysis team: objective modeling, predictive modeling, and prescriptive modeling.
3. The purpose of descriptive modeling is to excavate and review the current situation. NYT Reader Scope is the application service of descriptive modeling. From readers' preferences, places of stay, etc., NYT Reader Scope can understand readers' faces.
4. Predictive modeling can continue the function of descriptive modeling, learn from it, test it, and make the data model have predictive ability. For example: The Project Feels project is that The New York Times uses a machine learning model to analyze readers' reading emotions. After collecting and analyzing emotional data, it recommends articles that are suitable for readers.
5. Prescriptive modeling was described as very good in the speech, which can formulate corresponding implementation plans for future goals, similar to the extension function of predictive modeling. Therefore, I guess that cultivating readers' reading preferences is one of the possible applications.

MORE

• Data Science at The New York Times
  A full briefing of this talk. I haven't found the latest version, but this one was uploaded in January 2019, so the content is not much different.
• data: past, present and future
  CW's data analysis class at Columbia University, the 2020 lecture notes and code exercises are all placed on GitHub for free access. I decided to finish his class this semester.
• @chrishwiggins
  CW's Twitter account, although it hasn't been updated, still has to catch up with a pilgrimage mentality~
• chrishwiggins
  CW's GitHub account is updated very diligently (because Columbia University textbooks are placed on GitHub and updated regularly every week)

Data visualization: hard vs easy

Processing that saves the handicap handicapped

Simple first. Danish digital artist Tim Rodenbröker offers a free online course to teach everyone to learn Processing from scratch because everyone is staying at home to prevent the spread of the virus. Processing is a Java-based visualization tool. Because it is easy to learn and easy to use, it is a favorite tool for many artists and interaction designers. If you are familiar with it, you can create complex visual animations in a short time. At present, the courses are still being updated, but most of them already have teaching videos. It took me half a day to get familiar with Processing through the teaching videos. I believe that those with programming experience will definitely get started sooner. (Looking at the animation that I made by myself, it is super fulfilling!)

MORE

• Creative Coding Essentials
  Tim's Processing instructional video has a total of 27 lessons. According to what I've watched before, there are only about 5 missing.
• tim rodenbröker creative coding
  Tim's YouTube channel is really passionate about teaching, and he is worthy of being an artist + designer. The video is so beautiful that it surpasses other programming videos.
• @timrodenbroeker
  Tim's Twitter account, such a prolific art designer, was of course there to see the visualizations.
• Processing.py syntax reference
  If you like to write Python like me, I recommend Processing.py, download the Python package in Processing, and change it to Python mode on the edit page, you can type a lot less { }

Rediscover D3.js

Again it is difficult. As long as you are in contact with data visualization, you should probably touch some JavaScript-based D3.js. Since I accidentally read the classic article The Hitchhiker's Guide to d3.js , the content clearly explained the D3.js learning path and goals, which rekindled my motivation to improve D3.js skills. Because the content provides so many resources, I can't digest them all in a while, so I decided to start with the most economical path: FreeCodeCamp. Interactive programming teaching, and after completion, you can get a data visualization certificate, well, it's perfect. I finished it in the early morning of last Sunday, leaving the last project to be implemented.

MORE

• FreeCodeCamp Data Visualization
  This course is divided into three parts, the first part is the basic grammar, the second part is the operation of JSON and API, and the last part is to implement various charts by yourself.
• @mbostock
  Mike Bostock is the father of D3.js, whose passion for data visualization led to the creation of Observable, a JavaScript version of Jupyter Notebook.
• List of Graphics People in Newsrooms and Awesome Interactive Journalism
  These two lists are the twitter accounts and data news works of various data visualization reporters. Most of them are presented in D3.js. Just look at the power of D3.js.

This article was also published in the Tinker's Backchannel newsletter that I run.

Tinker's Backchannel shares weekly content on data analysis, visualization, news media, and interactive design. If you are also interested in these, please subscribe and communicate.