Todays Top Picks: A digest of articles and papers
Check out some of the notable things we read over the past week:
Bayesian Solutions for the Factor Zoo: We Just Ran Two Quadrillion Models
With data availability and computing power at an all time high, people are able to use brute force methods to analyze incredible swaths of information. To this effect, researchers from the UK constructed and ran over two quadrillion models in an attempt to determine which factors are truly predictive of stock prices. This one study might not change the game, but this depth of analysis could be a way to find an edge:
The 4 Hottest Trends in Data Science for 2020
Interest in data science has quadrupled over the last five years but the data science pipeline still involves manual work for data scientists. A number of those time-consuming tasks are being automated. With the ever increasing amount of data being generated, data privacy and security will become increasingly important. Cloud computing will play an important role to access the computing power necessary to process the volumes of data. Natural language processing will continue improving its ability to extract information from larger bodies of text.
AutoML for Time Series Forecasting
A competition winning approach and lessons learned on using AutoML for time series forecasting with the aim of handling new datasets without human intervention.