How do you make Amtrak safer?

Tomorrow (actually, already today) marks the first class of this Spring semester BUS 351 at Emory Goizueta Business School. But before we take off for a new exciting semester, I would like to comment on some great work done by students in the last year’s cohort. The accolade goes to GBS students Gobie Kumarasmy, Jamie Landman, Shannon Lin, Jeff Meng, and Aaron Weiner for their project on the safety of Amtrak. Basically, they asked the question which is in the title – how would you make Amtrak safer?

Amtrak is the largest publicly funded US passenger rail operator, so the debate about spending there is rather intense. First, let us understand how many accidents happened with Amtrak trains over almost 40 years, from 1973 to 2012. Accidents can be of two types: those involving rail equipment (for example, locomotive breakdowns), and those involving crossings with highways (for example, car crashes into a train). Results are reproduced with the authors’ permission.

You can see that highway accidents are mainly confined to the coastal areas, as expected. Somewhat surprisingly, southwestern states have a lot of equipment accidents. So what drives the accident rate?

For the highway-rail accidents the key drivers are weather and the presence of lights at the crossing, whereas visibility and presence of the warning signs are not. We cannot do much about weather, but it seems that lighting more of rail-highway crossings can help.

The situation gets more interesting if you look at the rail accidents. First of all, one might think that the age of rolling stock is a significant predictor of the accidents. Somewhat surprisingly, it is not. Neither is the train speed. Extreme temperature, it turns out, is associated with a higher rate of rail accidents, but it is not the only one. Amtrak, it turns out, has a history of drug-related problems. Look at this data:

Accident rate seems to decrease when more drug testing is done, but what is really interesting that simply bringing attention to this problem (2012 inspector’s report) has roughly equivalent effect.

All this data gives food for thought about importance of work culture for successful operations. Sometimes one needs to look beyond obvious solutions (such as invest in new locomotives), to improve the situation. This can be cheaper and more effective.

Where does the Big Data come from?

The Big Data is getting big, but where does it come from? And what does it mean for people? The recent WSJ article offers a nice interactive feature on the subject.

There are some obvious sources like social networks and point-of-sale data, but some are less so. For example, that multimedia/navigation/emergency help system in your car can be routinely tracking your driving habits, and, possibly, sharing the data with insurance companies. Another recent (and controversial) source of the location data is the license plate tracking systems, implemented both by police and private companies.