Matt Meshulam's Brain Dump

As my tires were spinning out in the snow for the dozenth time, I wondered if there is a significant reduction in fuel efficiency in the winter months. It seems like there should be a pretty sizeable effect: When it’s cold, people idle their cars longer, they drive slower, and any snow or ice makes for a less-than-optimal tire-to-road interface. Since I’ve been doing quite a bit of nothing since I’ve been home for break, I took it upon myself to find out the answer, with a little help from some of my favorite federal government agencies.

The question I focused on was, “is there a change in average miles-per-gallon as temperature changes?” I could have looked against precipitation levels, or something else, but for now I figured to stay with just temperature. I picked a handful of U.S. states over a range of climates to partially control for any seasonal effects.

Data sources

I got monthly state-by-state breakdowns of petroleum product sales from the Department of Energy. The number I decided to use was the total for motor gasoline, which includes unleaded, premium, etc. but not diesel fuel. I chose to omit diesel since it is used for much more than just trucks, and it would be unreasonable to expect an accurate absolute “miles per gallon” number anyway. My goal was to see how personal auto efficiency changes with temperature, and as long as there are no drastic changes in the number of trucks and other diesel-powered vehicles on the road, the trends will still come through.

Now that we took care of the “per gallon” part, we need the “miles driven” that goes up in the numerator. For that I used Table 5 of the Federal Highway Administration’s Traffic Volume Trends data. This data uses a sensor network across the interstate highway system as well as some dubious statistical witchcraft to estimate how many million miles were driven on all roads in each state each month. As far as I could tell, this includes commercial traffic like those diesel trucks I mentioned above, but again we’re not going after a literal miles-per-gallon number so this won’t be reflected in the trends. I didn’t get the best feeling from this data, since there were often some pretty drastic month-to-month changes in miles driven, but I take whatever data I can get. The numbers I used for each month were taken from the following year’s report, which goes back and adjusts the numbers to agree with other, more reliable, highway data. I’m not sure it was the right decision, but they weren’t too different so I’m not going to sweat it.

Finally, I got monthly mean average temperature data from Weather Underground (not to be confused with the other Weather Underground). This takes the average of the high and low temperatures each day of the month and averages them. I figured this was the simplest measure of the temperature over an entire month. The temperatures for each state were taken near the state’s center of population, but I tried to stay away from large, climatologically diverse states like California.

The states I used were: Minnesota, Wisconsin, Ohio, and Georgia. This group reflects a range of climates to control for strictly seasonal effects. Furthermore, the states are relatively uniform in climate (no mountainous states were chosen for this reason).

Results

About 24 months of data were taken for each state (some months had incomplete data). An estimated miles per gallon was calculated and the median for each state was taken as a baseline measure. For each month, the deviation from the state median efficiency was plotted against its temperature:

Although the variance is somewhat high due to the approximations taken, there is a significant downward trend with decreasing temperature below about 40 degrees. As it goes from 40 to 20 degrees Fahrenheit, you lose about two miles per gallon, on the order of 10%. The quadratic trendline gave by far the best fit, with an R-squared of almost exactly 0.5. Note that the trendline indicates a slight dropoff in fuel efficiency with very high temperatures, but I’d need to throw some southwestern states in to see if that trend holds up.

No posts for about a week and a half; I’ll be in Israel.

I built this up this morning, it was a great way to break in my new cordless Dremel. My Sansa e200 series MP3 player only charges via USB, so I made a standalone charger with a female USB jack. It’s just an old 5-volt cell phone charger, but I replaced the cord with a jack from a USB extender cable. A little epoxy and electrical tape, and voila!

Naturally, the red and black wires on the USB connector go to the +5 and 0-volt connections in the wall wart. I spent some time trying to figure out when needed to be done with the white and green data wires. Some people online said they needed to be held at some intermediate voltage, but didn’t really know why. I found some specification for standalone USB chargers that said to short the two data wires together, so that’s what I wound up doing. I have a feeling it doesn’t really matter, but that worked for me.

I was hoping I’d be able to use this with my Motorola phone as well, since its charger has a USB-b connector, but I didn’t realize that the USB-b plug has a fifth “ID” pin so that all the manufacturers can pick a different resistance, rendering all the different brands of chargers incompatible. I think I could still use my device, but I’d also need to modify a USB cable, and seeing how my current charger works just fine I’ll hold off on that for now.

The Onion AV Club has published its annual list of bad band names. I really recommend scrolling through as much of the six pages as you can bear, but here are some my favorites:

• Touched by a Janitor
• The Tao Jones
• Abracastabya
• Insects with Tits
• Eskimo Kisses for Mommy
• Piss Pissedofferson
• Sans Pants! Ska Band!
• Sigmund Droid
• Nacho Cheese And The Headgear Honeys
• Elf Lettuce
• E=MC Hammer

On a recommendation from XKCD, I picked up a copy of The Visual Display of Quantitative Information by Edward R. Tufte. So far it’s been a surprisingly interesting read, and I recommend it to anyone who likes graphs.

It also features an image that was discussed in my technical writing class last year. It is Charles Minard’s graphic documenting Napoleon’s disastrous attempt at invading Russia in 1812:

Now before your eyes glaze over, hear me out on this. First, I highly recommend clicking the image for a larger version courtesy of Wikipedia. This graphic tells a story far more succinctly than any history book. The top part is a map of Russian territory. The width of the tan line represents the number of troops as they march toward Moscow. The French army started its campaign with 422,000 troops as it crossed the Niemen river. Twenty-two thousand shortly split off to guard the rear as the rest of the army marched through what is now Lithuania. Another contingent separated from the main group to guard the flank. By the time Napoleon reached Moscow, he was left with only one quarter of his original forces.

The retreat, which started in October, is where the graph gets really interesting. In addition to the black line plotting the location and number of the French troops, the scale at the bottom displays dates and temperatures as the army trudged home through the Russian winter. In early December, as the army was nearing the end of its reatreat, temperatures dropped to -30 degrees Reaumur, about -35 Fahrenheit. By the time Napoleon crossed back over the Niemen, his nearly half-million-strong army was reduced to 10,000 troops.

I’m in no way a history geek, but I’m still fascinated by the story Minard tells using only quantitative facts. In his book, Tufte points out that six variables are expressed in the graphic: two dimensions of space, number of troops, direction of troop movement, time, and temperature. He declares, “It may well be the best statistical graphic ever drawn.” I think I’ve got to agree.

I know it’s been a while since I rapped at ya, but finals season was keeping me busier than I expected. Now I’m back in Chicago for a few weeks so I hope to make up for my recent lack of posts. I’m going to warm up with a few short bloglets:

• As a followup to my previous Megabus story, as I arrived at the St. Louis stop to travel to Chicago, I was greeted by a Megabus dispatcher. She couldn’t have been more than a senior in high school, but there she was with her clipboard, checking people off. It made me feel old. But I have to hand it to Megabus, they’ve discovered a way to grow their operation while simultaneously reinforcing their image as a no-frills, DIY establishment. Child labor, is there anything it can’t do?
• Speaking of transportation, picking up Eve at the airport turned into a 4.5-hour trip because of a few inches of snow. I blame Rod Blagojevich.
• Speaking of politics, I can’t say I’m super thrilled about the reports that Tom Vilsack has been chosen as the next Secretary of Agriculture. I’m sure we’ll be hearing stories all day tomorrow about how he’s a perfect choice because he’s an environmentalist and still has good relationships with agrobusiness, but to me, the symbolism of choosing the former governor of Iowa as agriculture secretary leans a little more toward “more of the same” and less toward “change we can believe in.”
• Speaking of the Department of Agriculture, the NY Times had an op-ed a few days ago that called for the department to be renamed the “Department of Food.” It goes into how the idea of an agriculture department is outdated and how the needs of the people actually eating all the food should be better represented. So much for that.

A pop-up notification from Google Calendar:

Turn in timesheet (Matt Meshulam) is starting at 12:30pm in timesheet

Google uses some simple text parsing to attempt to determine the time and location of an event you enter. In this case, I typed “Turn in timesheet.” My intended meaning was “I must submit my timesheet.” Google interpreted my text as “You must go to some place called Timesheet, and when you get there, you have to turn.”

The singularity is not yet upon us.

Megabus Chicago, November 30, 2008. Click for full panoramic.

Yesterday I took the Megabus for the twelfth time. As one would expect on the Sunday after Thanksgiving, their Chicago hub was bustling. They ran three buses on the 9am route to St. Louis, and two each to Toledo and Detroit around the same time. When I got to the “station” — a folding table on Canal Street — I had my first feeling that perhaps their business model wouldn’t be able to scale up to a full-fledged national bus line. I’ll start out by ruining the suspense: My bus left only 20 minutes late, made up the time on the way, and got to St. Louis right on time.

Now that that’s out of the way, let’s get down to brass tacks. Megabus is cheap because it doesn’t have bus terminals. Megabus is popular because it’s so cheap. And when you try to herd 500 cold, hungover college students onto the correct buses, there will be something resembling a mild form of chaos.

A quick rundown of observations:

• Apart from the drivers, I saw only three Megabus representatives: one man checking passengers’ confirmation numbers, and two people directing the buses and helping to load luggage. None of them were wearing Megabus clothing, and I didn’t even notice any I.D. badges.
• At the check-in table, each rider was given a paper ticket indicating which bus they should board. This was a good move, as it prevented everybody from mobbing the first bus heading to their destination. And the buses all pulled away within about 10 minutes of each other.
• Possibly the biggest impedance of efficiency was that at a given time, only two buses were pulled up to the curb loading passengers. It appeared that the whole block was intended to be used for bus loading, so I’m not sure whose decision it was to only have two parked at a time.
• I overheard quite a bit of confusion from passengers about where to check in, which bus was theirs, et cetera, but there was always a Megabus staffer hollering sufficiently loudly for information to disseminate. I know they enjoy their distinction of being a no-frills bus line, but I think a megaphone would be a fantastic investment.

Megabus and its parent company, Coach USA, should be commended for maintaining a semblance of order in a situation which could have easily devolved into complete anarchy. According to a press release, Megabus’s revenue has more than tripled in the past year, but I don’t see how they would be able to manage twice as many passengers next Thanksgiving without some actual infrastructure. It’s apparent that there’s a huge market for cheap inter-city bus transit, especially with the economy and all, so I hope Megabus will be able to expand their service while keeping overhead costs at practically zero. And I hope that everybody gets to know the thrill of snagging the elusive $1 fare, which will be a case study for introductory economics classes for decades to come.