Limitations
- As they mentioned in their paper, by aggregating the behaviors of many people, the algorithm itself may be prone towards a "majority" bias consequently misrepresenting or hiding behaviors in the minority.
- Moreover, their data are based on a limited sample of check-ins shared on Twitter and are therefore biased towards the types of places that people typically want to publicly share. The demographic of foursquare users, which is usually characterized as young professionals in the ages between 25 and 35, owners of smartphones and urban residents can also influence the results.
Things needing improvements
- Color Codes for Livehoods: When I looked at the visualization at first time, I thought the Livehoods with same colors must be related with each other. But it turned out that the colors were likely to be randomly picked for Livehoods, after I clicked the "Related" tag in the control panel. So, I think it would be much better if they can have a nicer color codes.
- Timeline: The visualization seems to not have an option for user to select a time range to generate the Livehoods dynamically. I think it would be very helpful for user to analyze how the Livehoods in their cities change dynamically according to time to understand their cities in a much better way.