Unraveling NFL Win Probability Graphs to Find the Best Games Nov 13, 2021 How do you determine how good a game is from its win probability graph? You could compute the Excitement Index by adding up all the changes, or you could take a visual approach and “pull” all the slack out of the chart.
Stop Taking L's When Walking to Your Destination Jun 27, 2021 We’re often directed to get from one point in a grid to another by taking a single turn and otherwise walking in two long segments. But the Walk/Don’t Walk signs that dictate our travel make this approach very inefficent, and motivate a better strategy that may save you a few minutes.
Estimating When CitiBike Rolled Out E-Bikes Using Change Point Detection Mar 21, 2021 If electric bikes allow their riders to travel at greater speeds, will there be a corresponding dip in average trip duration between bikeshare docks that we can find in the data? We try to answer that question by applying a change point model to a notoriously difficult uphill route.
Visualizing How a Kernel Draws a Smooth Line Feb 13, 2021 How does software like Excel or R trace a smooth line through a series of points? It’s easier to understand by visualizing how new values are created as a weighted mean of the existing data.
Gradient Ascent: Finding Uphill Routes for my Long Walks Jan 23, 2021 An old back injury makes it so that walking uphill is much more comfortable for me than walking downhill. A programmatic search for uphill routes reveals how strict constraints and “greedy” decisions can lead to sub-optimal results.
Avoiding Variance on my Morning Commute Jan 10, 2021 Sacrificing value for certainty is a common tradeoff in life. Is this what I was (accidentally) doing when I started walking the first part of my daily commute instead of taking the subway? Some historical MTA data can help answer that question.
Pushing Google's Autocorrect to the Limit Dec 26, 2020 Everyone makes errors sometimes when hastily typing a query into a search bar, and most of the time Google is able to recognize your mistake and offer a suggestion. But what is this algorithm’s breaking point? How badly can you mis-type a word before even Google doesn’t know what you mean anymore?