Synopsis: Thomas Davenport’s ideas on how to work with or manage Data Scientists.
For the uninitiated, “quant” comes from “quantitative analyst,” someone who works with data. As far as I can tell it originated back in the 80’s on Wall Street, with a group of really bright people who figured out how to apply advanced math and statistics to the prediction of stock and bond price movement.
Davenport’s intention with this book is not to turn anyone into a data scientist, but rather to introduce the idea of using analytics to solve business problems. In a nutshell, the plan goes like this:
- Define the problem. Are you trying to improve your revenue stream, tighten up your supply chain, or something else? It’s incredible how many companies have built out a Big Data infrastructure without first defining what they want to accomplish with the data.
- Review previous findings. This is a step which is all too often left out of other books. Frequently the problem has been examined before, whether in your own company or by someone else who published their results.
- Modeling. What most people think of when discussing data science. Part and parcel of this step is determining what data is needed and how to acquire it.
- Collecting (and cleaning) data.
- Analyzing the data. Basically, running the data through the model and ensuring the results make sense, to include making sure you haven’t overfit the model.
- Presentation of results. Not to minimize the importance of the other steps, but this is probably the most important. If you can’t convey the meaning and value of the data to decision makers, it really doesn’t matter how great your model is or how ground-shaking the results may be.
Davenport spends the bulk of his time going over these steps in more detail, using real-life stories from companies like Cisco, Cigna, and Renaissance Technologies. The examples are pertinent and his style is eminently readable.
If you are starting from zero on the path to becoming a data scientist, or if you are a manager wanting to work with data scientists to solve your business problems, I can think of no better book to recommend.