Tech Economics Newsletter
What's new from economist Andrew Chamberlain, Ph.D.
I'm here with my latest update from the world of tech, economics and data science:
Reproducibility Crisis in Machine Learning? One of my daily struggles as a researcher in the tech world is lousy documentation -- code without comments, tables without documentation, and a wiki that's constantly out of date. How can data science teams stand on the shoulders of giants if we can't replicate each other's results? Here are some good ideas about making machine learning more transparent and reproducible.
Awesome Web Apps with R Shiny: Tableau is great for data visualizations. But it's slow, and proprietary licenses mean it's hard to share internal dashboards publicly. Here's a terrific open-source alternative -- R Shiny. Anything you can build in R can be transformed into an elegant web app with a few lines of code and a hosting account with shinyapps.io. Here's a fun example from our team.
The Real Data Scientists of Glassdoor: What's it really like to run a machine learning team at Glassdoor? Here's a terrific interview with Ling Cheng, one of our data science leads at Glassdoor and a true innovator at the company. She's brilliant and one of the people I've learned the most from in the data science world.
Monopsony and Online Job Postings: What can we learn from the millions of online jobs floating around the internet today? These data are messy, but they can provide powerful insights about real-time hiring trends in the economy. Here's a terrific new paper using jobs data from the good folks at Burning Glass Technologies to measure labor market concentration for jobs. Bottom line: Labor markets today are a lot more concentrated than we thought.
How to Fix Online Reviews: We wrote a short piece for Harvard Business Review on our new NBER working paper on how incentives can reduce bias in online reviews. Turns out prompting people with simple, inexpensive pro-social messages goes a long way toward getting better online reviews.
Write Code Like a Boss: A big difference between academic and data science research is whether you write code like a story, or build it like a machine. For research, you treat projects like a story and write a paper when you're done. In data science, you often build models that go into production like a machine, plugging it into a huge ecosystem of other code. Here's a smart article about writing better production level code -- sometime I'm trying to do better in 2018.
Come Write Papers With Me: I'm hiring an Economist / Data Scientist and a Research Analyst at Glassdoor's headquarters just north of San Francisco. My dream job in college was to run my own think tank. After 15 years I finally did it -- and now we're looking to grow. Let me know if you can recommend someone great!
Thanks again for staying in touch with me -- and best of luck in the coming month.
Andrew Chamberlain, Ph.D.