This is an attempt at building out a series of introductory labs in KDD, or Knowledge Discovery in Databases. The words "data science" are thrown around a lot, but here it is in full:
- Data set selection
- Data modeling and architecture
- Pre-processing, or "cleaning," data
- Data transformation, or "ETL" (extract-transform-load)
- Statistical modeling
- Data mining
- Data analysis or interpretation
- Predictive analysis and "CEP" (complex-event-processing)
Many of these are mostly applied skills, some or conceptual, but all are a mix, and each is intertwined with the others. That's why presenting them in a single course of work, themed around a single domain (or "problem"), and with a unified, clear data set, makes it a little easier to understand the process as a whole.
See the Voter Lists README and the Demographics README.
Imagine it's January 1, 2016. You recieve a call from a friend of an old friend, whose heard that you are a "computer person." The caller identifies themselves as representing a candidate for national office. The candidate is running an insurgent campaign that's had enormous and surprising success, but hasn't had the resources or staff to create a presence in the crucial state of Ohio. For starters, they are going to need a "data game," someone to run the numbers, build the models, deploy some applications, and advise on strategy.
You are being offered that job. You're unprepared, but hey, you gotta learn somehow… Politics is all about "on the job training," right?
…