From CS294-10 Visualization Sp11
- Michael Porath
This visualization project is a contribution to a bigger research project that focuses on behavior change in the domain of transportation. Personal transportation has a big impact on the environment and individuals. For context, here are some data points:
- Congestion on roads leads to a waste of fuel and time
- 5.7 gallons of fuel are wasted annually due to congestion on roads.
- $67.5 billion is lost in terms of productivity. This is almost 0.7% of the US GDP.
- Personal transportation contributes to
- 26% of CO2 emissions in the average US household
- 2/3 of oil consumption in the average US household
The larger goal of the research is to influence people to drive less and use more sustainable modes of transportation.
For this class project, I will be focusing on giving users feedback on their driving behavior. I have access to a database that collects discrete data points about people's driving behavior. This information includes the user's GPS location, sample times, speed and acceleration. The data has is being collected with a mobile app that runs in the background. The goal is to correlate that data with other data layers that impact driving behavior, such as road congestion and accidents and crime. The visualization ultimately should give feedback on a personal level and ultimately lead to behavioral change.
Initial Problem Presentation
- Link to slides: File:MP Proposal Slides.pdf
Midpoint Design Discussion
- Link to slides here
Jvoytek - Apr 04, 2011 03:12:36 pm
- Very intuitive mapping of latitude/longitude and efficiency
- I like the brushing and linking between the different views
- The addition of external data with congestion could be very useful
Brandon Liu - Apr 04, 2011 03:13:47 pm
I like the visualization of velocity and acceleration during each trip, but would it also be possible to visually compare how long each trip took? If the goal is to change user behavior, it would be helpful to know how changing their driving habits influences commute time, in order to weigh the cost/benefits of different routes.
Julian Limon - Apr 04, 2011 07:41:03 pm
I like the concept a lot and I had a few thoughts during your presentation:
- If the goal is to use this visualization to influence people and change behavior, I believe that it could also be matched with other data to make the argument more compelling. For example, MPG, time, or cost of the trip could be compared to public transportation or to carpool lanes (i.e., how much could the person could have saved had she had more people in her car or had she taken public transportation)
- As Thomas mentioned, adding the time component could be a great opportunity to allow people to analyse their driving patterns. If they realize that they save gas or time when they leave at a different time, this could make the change easier.
Sally Ahn - Apr 04, 2011 10:21:14 pm
The goal of this visualization--changing people's behavior--is quite interesting. One might want to consider when and how this visualization should be displayed (e.g. mobile phone, a car's navigation system, etc.) to create the most impact on a person's behavior, and try to optimize the visual variables for such devices.
Michael Cohen - Apr 04, 2011 11:47:54 pm
Julian beat me to the punch a bit, but I'll add that in addition to public transit, carpool lanes, and times of day, it would be useful to have a comparison to different driving routes. Perhaps there are back roads that appear slower at "normal" speeds and thus aren't suggested by online mapping tools or GPSes, but are, in fact, faster during rush hour. (Or, if not faster, perhaps the same speed but steadier in a way that avoids acceleration/deceleration.) I really think that some form of comparison is key -- if you want to change behavior, it's not enough to show people that their current commute stinks (most commuters are well aware of that!). You also need to show them at least one viable alternative.
Great mission with this project, by the way. I'm looking forward to the result!
Siamak Faridani 01:13, 5 April 2011
I find the idea of using visualization as a means to behavior change very compelling. I am wondering if applications like UbiFit exist for fuel consumption. One thing that might be also interesting to look at might be cases where there is no linear relationship between the visualization elements and data (something like Ford's green dashboard where number of flowers are a representative of the fuel consumption) Perhaps you can use similar ideas to show congestions. Great draft by the way.
Matthew Can - Apr 05, 2011 02:41:37 am
From the presentation, I got that the purpose of the project is to provide users feedback on driving behavior so that they can use it to change their behavior. But it wasn't clear to me what kinds of behavior in particular you want to change. I think it would help if you tie this in to the design you choose for your visualization. I thought you did a great job addressing the challenges you expect to face and storyboarding the interface you plan to build. I realize that a full blown evaluation of your interface would be outside the scope of this class, but it might be a good idea to design an evaluation, even if you don't carry it out.
Karl He - Apr 05, 2011 03:24:14 am
I like the way you plan to encode the speed with the line width. It may be beneficial to somehow combine the acceleration and speed visuals, although I don't have a good idea of how to do so. This would make it easier to distinguish simply slow driving to frequent-acceleration which is the real problem.
Michael Hsueh - Apr 05, 2011 02:58:23 pm
I really like this idea and I think there is much benefit to be had in visualizing driving behavior. I also like others' comments about being able to provide alternative routes for getting from point A to point B. One of your slides shows the aggregate data collected over multiple routes. Extrapolating this idea a bit, perhaps a way of identifying better routes is to consider data collected via other routes. This data might or might not come from the same user. I know your project is limited to the intra-personal case, but down the road, incorporating inter-personal data for the purpose of constructing a more "instructive" data set might be helpful.
Saung Li - Apr 05, 2011 07:49:58 pm
I think your idea of showing velocity and acceleration below the map is great, as we would like to reduce breaking and accelerating too much. For the purposes of changing behavior though, I'm not sure if this is enough. Users may want to know how much gas they have consumed during a driving trip, and be able to compare this with how much gas and money they would have saved if they took a more efficient route at that time of day. Money may be a bigger influence in people's driving than just seeing velocity and acceleration.
Manas Mittal- Apr 05, 2011 09:03:36 pm
I like the focus on behavior change, but even independently, this project has the potential to be very interesting and useful. I am assuming that you have timestamped data and can use it to build time-dependent traffic-encoded maps. Perhaps the user can use a time slider and see the effect of a given road (averaged over many days). This would give people an idea of when to drive (so, it would lead to a indirect behavior change - people will hopefully drive less at the time of congestion. That also seems to be a good thing).
In related work, it would be worthwhile to check out the Cartel Project - http://cartel.csail.mit.edu/doku.php?id=CarTel Sam Madden, the main professor behind the project is at a Sabbatical this year at Berkeley (he sits in the RadLab) so you may want to talk to him. He might be willing to share the data with you in case you want a head start with the visualization.
David Wong - Apr 06, 2011 01:18:39 am
I wanted to reiterate that the feedback on driving behavior to change behavior should reflect some of the behavior change strategies that the project is taking. So if one strategy is through goal setting, the visualization can reflect how close the user is to reaching certain goals (eg. 20 less miles on the car per month). Even outside of behavior change strategy, your visualization can focus on well established goals for drivers, eg save time and money on gas (maybe even encouraging drivers to possibly take walking routes if the distance is short). With a clear idea of what the message the visualization should give and an idea of how that informs behavior change, the visualization should come together quite nicely!
Dan - Apr 06, 2011 10:23:07 am
I like how the big picture is to improve behavior of individuals to benefit the sustainability and mother earth. Good way to be on the right side of history. I wonder how the psychology of visualization plays into chanigng behavior. Interesting topic. Also, I think the mapping for lat/lng was done well.