30 Days of Data Visualization

In a past life, I had a bit of a reputation for data visualization – in fact, one of my biggest pet peeves was watching someone make a decision based on their perception of information, rather than using real data. It’s shocking how often this is done, and it’s especially horrible considering any basic study of human psychology shows that we trick ourselves constantly when it comes to processing information.

Soap box aside, I have been trying to gather certain pieces of data about my travels over time. In some situations I have massively dropped the ball: I stopped gathering gas price and consumption data when I started using a spare gas can, something I really regret. In others, I’ve relied on technology that has failed at times: my detailed GPS datalogger locked up and was unusable for three days until the battery died.

In spite of this, I have gathered some interesting data that I thought I’d share, as well as my analysis of said data where appropriate. Some of this has been gathered by hand in a notebook, while others are from electronic sources. I’ve denoted the source where appropriate.

On April 16, 2011 I left Lima Peru. On May 13, 2011, I arrived in San Martin de Los Andes with a broken front wheel for repairs. All data below is from this period, of the route travelled below (GPS data provided by my SPOT Satellite Tracker and is thus only accurate on a macro scale):

The graph above is coded by elevation – as you can see, I spent a large period of time at high elevation on the altiplano in the Peruvian Andes, while most of my time in Chile was spent near to sea level in the desert on or near the Pacific coast. In Argentina, my time has been spent in the shadow of the Andes, meaning slightly higher than normal elevation but rarely the true mountain elevation.

This is illustrated below in a graph showing specifically average elevation over time (note that I actually hit peak points of elevation considerably higher than shown here, however these were not logged by my SPOT since it only reports current location at long intervals):


This clearly illustrates both the majesty of the Andes mountain range as well as the amazing altitude of the Peruvian altiplano. Consider this: the highest point in the continental US is 14,505 feet, and the highest road in the lower 48 reaches 14,120 feet, while being mostly below that. Even more interesting is that the highest town in the US which has permanent residents is at only 10,355 feet while many large cities in Peru are much higher – though of course this is interesting as perspective only, since climate differences due to latitude come into play as to why we don’t reside so high in the US permanently.

With the caveat that the data below is also informational only since it is not captured at a frequent enough rate to provide a high level of accuracy, here is a graph showing climb rates over time as captured by my SPOT:


If you ignore the outlier for ascension on 4/30 (this was when I climbed the insanely high dirt road towards Tacna, spending the night at over 15,000ft), it shows a very interesting trend: one does not only climb or descend when riding through the mountains, but rather the roads tend to do both over time.

In fact, in general, you do both at around the same rates over time when dealing with stable areas – in other words, aside from variances such as climbing up to or down from the altiplano, in general your climbing and descending tends to cancel each other out over time. In a way, this is an obvious conclusion, but it is neat to see data to back it up.

Of course, another interesting piece of data – also obvious – shown by the above graph is that in general rates of descent are higher than rates ascent, however this is not always by a considerable amount! Certainly not as much as I would expect when looking at more granular data. My conclusion is that in general caution dictates a rate of descent at typically less than full throttle/speed, while ascent is always done at the maximum speed sustainable.

There’s a similar piece of interesting companion data shown by the graph below, which is segmented into distance covered (“horizontally” on a map) during each day of travel:

If you compare this by day to the above graphs of elevation (4/18 to 4/29 being high elevation), you can see that in general the amount of time spent traveling at high elevations was the same as at low elevations (the horizontal distance of the lines on the graph above), while also in general the total distance traveled on a map (the vertical distance of the lines above) is less at high elevation. Conclusion: Traveling in mountains is less efficient than traveling on flat empty plains. Seems obvious, right?

But wait – is this actually true? The data above is based on the horizontal distance between track points as tracked via my satellite tracker and does not necessarily take into account the actual distance travelled as a road in the mountains rarely takes the shortest paths between two points. This is illustrated by the fact that the total distance tagged by the satellite track is approximately ~500 miles less overall than the distance tagged by my odometer (and my GPS datalogger which logs very few seconds, rather than at many-minute intervals).

The graph below shows my manually tracked mileage over time, as written in my notebook and calculated based on the odometer readings at the start and end of the day:

That shows a clear trend of increasing average travel distance over time (as illustrated by the trend line), which may corroborate the above graph or simply point to an increase in how hardcore my attitude about travel is – we can’t say which without adding a critical factor, specifically: time.

How much time I spend traveling each day obviously has a direct impact on the distance travelled, possibly as much as the elevation and thus probably curviness of the road. Let’s add that in:

This one I think is fascinating – it clearly shows that over time, I have trended towards averaging around the same number of hours of travel each day, however I have clearly trended towards a noticeable increase in average rate of speed (taken by dividing distance travelled in a day by hours travelled in the backend).

I think at this point it’s a fair conclusion to draw that I am making much better time in Chile and now Argentina than I was in Peru. Whether this is due to elevation or the nature of the roads (paved vs dirt) I can’t clearly state, as I don’t have the data to show how much mileage was in what kind of road condition… but definitely, if you’re planning on repeating my route, you can assume you will travel faster on the middle part of Ruta 40 in Argentina than you will on the altiplano in Peru.

On a completely separate note, I have not been tracking my expenses as I would like, mostly due to stress. I do, however, have one interesting statistic I can share before I stopped tracking it:

From April 16 to April 30, while traveling from Lima to Tacna Peru, I travelled a total of 2354(1471 miles) with gas data logged. During this time, I consumed 28.25 gallons of fuel at a cost of s/382 PEN ($139USD), an average of s/13 per gallon ($4.9USD per gallon). Total overall gas mileage during this time was thus 52MPG at a return of 10.6 miles per USD.

I wish I had written down all the details in Chile and Argentina, because it is much, much higher – $7/gallon or more. Eep – maybe it’s good I don’t have that data.

(please note that as always I didn’t proofread this because I am lazy and drunk half a bottle of wine while I was fidding around with the data, so I apologize if somewhere you see something that doesn’t make sense… it’s just for fun after all, I’m not making a ten million dollar purchase suggestion based on it or something!)