Lognormality > Machine Learning: From Data to Decisions
18 April 2020 | | categories: school | blog archive
I recently completed the MIT Professional Education course Machine Learning: From Data to Decisions.
This course provides an overview of machine learning and related concepts at the “mid-level” (not quite high-level, but not too technical, either).
It was a good experience that got me thinking about my work in some new and interesting ways, and I’m looking forward to digging into some of the covered topics in more detail.
Key Takeaways
I’ve worked in “analytics,” for lack of a better word, for over fifteen years, but I’ve never seen such a clear framework for how to approach a problem as this:
- Understand data
- Make predictions
- Make decisions under uncertainty
- Determine causal inference
Most analysts I’ve known do variations of this, but many neglect to spend enough time on #1. Most of the work I’ve done has been in #1 (visualization, primarily) and #4 (hypothesis testing and time series forecasting), though I’ve done #2 as well. However, the discussion of step 3 will be especially helpful to me in the future, hence the bold text in the list above.