If you can’t do enough sport the next best way to burn calories is through incidental activity: walking to the shops, the local cafe, visiting friends, or whatever. I was interested in how much incidental exercise you can really do, and is it enough to make a difference? Can small trips made by active modes (walking and cycling) add up to big effects over the long run? For those Android phone users who have not opted out, Google is keeping track of every move and kindly reverse-geocoding the locations you visit, categorizing places and assigning modes of transport. This is really useful because we can then tell why a person is moving as well as how. The data is not perfect, and can be troublesome to analyse, but it still gives a pretty remarkable coverage and seems to capture those incidental trips. In fact the data is so good I wouldn’t dare use anyone else data, as it’s hugely personal. So we’re stuck with my data.
First, I looked at the number of trips I made of different journey lengths per year, excluding jogging or any other sport to focus on incidental exercise. Unsurprisingly I make far more shorter trips than longer trips (my chart has 1km bins for simplicity, but the data is to the nearest meter). Transport economists will note the data follows a power law distribution, and this concept underlies gravity models.
In the next chart we can see the distance this added up to over the year. Those sub-1000m trips, the yellow bar, came to over 400km. Most people can walk 1000m.
Everything under 3km comes to a whopping 1800km, which is like walking more than the length of the UK. I accept that 3km might be rather more than most folks want to walk.
The question then is how many of those small trips are done by active modes. The next chart shows the most common trip types for sub-1km trips. The categorization needs some work, but it seems that most of my short trips are to local shops and public transport stops. (Amenity is how I’ve groups local shops, restaurants and other people’s homes – this needs work).
If you live in a neighborhood where the shops are within a ten minutes walk, and where public transport is viable, it is inevitable that you will make these walking trips. The trouble is, much of the UK is sprawling low density housing, with limited walkable decent shops, and inadequate public transport. For much of the nation many of those short trips are done by car – people miss out on the free exercise because of the built environment.
The next chart shows the calories per day burned by walking these trips (assuming 30KCal per km walked – this is the low end assumption, and this varies between people). Walking all sub-2km trips burns 100 calories, which is a considerable dent into surplus calories – remember the body’s basal metabolic rate eats though at least 1500calories, so you only need to get through another 1000 by moving about and lifting things to get to 2500Kcals – a typical daily food consumption.
The timeline data analysis seems to suggest, in my case at least, that little errands are taking off 100kcal per day as long as they are walked.
Walking for half an hour burns 100 calories, reduces the risk of heart disease and strokes by 20-35%, reduces risk of diabetes by 20-40%, 20-30% lower Cancers, up to 80% fewer fractures and lower rates of depression and dementia (see this link for the full stats). There is strong evidence for these health benefits.
Only a multifaceted approach at the intersect of town planning, infrastructure, and behavioral physiology can deliver more walkable cities. The Mayors Transport Strategy published this week is leading the way in London, with a large chunk devoted to healthy streets. TfL have a spreadsheet tool, which scores developments for the attractiveness of streets, safety, pollution and access to public transport. A huge amount of work has been done for example in Walthamstow, where there seems to be cycle loops, Urbo hire bikes and directional signs on every street corner.
But Walthamstow has the density to support walkability – the high street shops and tube station are easily walkable from the dense terraced housing which typified the area. The real challenge is on the edge and outside the capital where large residential districts can be devoid of decent shops (a little costcutter selling overpriced ready meals is not adequate!) and regular trasport. Some research has been done on the impacts of food deserts in the UK (here,here). although I’ve not seen any maps yet. Food deserts have been mapped in the US and the situation in Atlanta is well documented. Much of the work on food deserts has focused on access to affordable fresh food in low income neighborhoods – this is a crucially important subject, but the connected issue that my timeline points to is walkability of good food.
I’ve not said so much about commuting. Clearly commuting by active modes is ideal, and implies a shorter commute leaving more time for healthier lifestyles. Unfortunately commutes are getting longer for a host of reasons. My timeline data suggests walking to the bus stop or train station is another major source of incidental activity.
If obesity is a key public health issue, alongside air pollution, it seems to me that free or very cheap public transport is a sensible idea. While some German and Belgian cities are leading on this, the UK government seems intent on cutting public transport subsidy.
Notes on the Google Timeline approach
I’ve extracted all the data from my Timeline, after Google has applied its geocoding to the location data. This means the dataset contains information such as addresses, type of place and mode of transport. Google doesn’t always allocate correctly. The dataset is incomplete because of periods out of signal or with the phone switched off.
To extract the data you need to scrape the google maps timeline search URL. I used Python Request library to grab KML files.
The data contains a wealth of information. Just scratching the surface of it. Below is a chart of rank distance versus distance, sowing that the distribution of trips follows a power law, confirming that gravity models can be a predictor of trip length and frequency.