Switch branches/tags
Nothing to show
Fetching contributors…
Cannot retrieve contributors at this time
87 lines (56 sloc) 3.5 KB

# Results

In this file I've collected some of the interesting results from this experiment. I'm sure there is more that can be done, but this is at least a start.

## Number of data points

Firstly lets check that we have generated a sufficient amount of data for the results to be meaningful. This can be done by simply plotting a bar-chart of the number of data points per country:

It appears that apart from a per countries, we have generated over 1000 data points for most of the countries. In the future analysis we should be wary of results from those countries with the smallest data count: GF (French Guiana), MK (Macedonie), and DO (Dominican Republic).

## Averaged speeds

The first and easiest question to ask the data is: "which country is the fastest?". Of course this comes with a huge number of caveats due to the data collection technique. Nevertheless taking a naive approach and just averaging the journey speeds grouped by country we can plot the averaged speeds in increasing order:

This paints a somewhat surprising picture: the fastest country (by this metric) is the US. This doesn't fit with many peoples expectations and we might guess this has something to do with the difference in size between the US and other countries.

The US should really be thought of as a collection of smaller countries (the individual states). If we don't do this, then picking any two points randomly by zip code in the US you will be more likely to pick a long-distance journey than picking any two points in another smaller country. The result is that almost all the journeys in the US are along interstate highways. I suspect that this causes a bias for faster journeys.

## Speed Profile

A more interesting way to investigate the data is to look at the "speeds profile" in a country. This described the distribution of journey speeds in that country and is calculated by binning all the individual journeys in a histogram.

For example lets compare a fast country Germany (DE) with a slower country such as Turkey (TR)

There is a marked difference in the distributions. Namely the slower country has a much longer tail in the increasing speeds direction while the faster countries speeds drops off sharply. This suggests the vehicles in the faster country are limited by a speed limit while in the slower country this is not the case.

### Some interesting cases

These two examples are quite interesting due to their double peaks