More commute data

After changing jobs, I revived my commute time data collector for my new commute. The end-result is much the same, e.g., Google Maps underestimates the commute time by an average of 20%. The range of underestimation (from the average time) was from 4% to 48%.

I programatically collected Google Maps estimates for a few days and averaged them, while manually collecting my actual commute times. The morning commute looks like this:

The evening commute, like this (don't ask me why Google spreadsheets chose a different display format):

The underlying data is here.