Think about forecasting results based on polls, how to visualize your results, and explain them. Now apply that to a flashy, real world example: the 2018 midterms. The core tutorial covers sampling, simulations, statistics, interactive visualization, and communicating results. Not to fear: this tutorial is ultra "hands-on" with every concept being immediately applied with NumPy, SciPy, and Pandas.
FiveThirtyEight produces many of the most accurate U.S. electoral forecast models, leveraging a wealth of polling, expert insights, and simulation. (See 2018’s House simulation.) In this talk, Matt Brems and Joseph Nelson break down the fundamentals of sampling, probability, and data visualization with Python so participants are able to recreate the methods like those seen in FiveThirtyEight. Finally, Matt and Joseph show how to deploy the results to an interactive website so that you can share your interactive visualizations and insights with the world.
The tutorial will focus on fundamentals in probability, sampling, and data visualization by writing clean, readable, and efficient code. Not to fear: this tutorial is “ultra hands-on” with every concept introduced being immediately applied with NumPy, SciPy, and Pandas. Visualizations will be created using Matplotlib. The emphasis will be on gradually understanding probabilistic methods, learning how to communicate these through visualization, then deploying results to a broad audience - all within the context of a real world case study to build from (the results of the 2018 midterm elections).
The tutorial is aimed at beginner and intermediate PyData participants. Ideally, attendees will have some experience with NumPy and Pandas. The core tutorial covers: