Dublin, As You’ve Never Seen It Before

To look at this image, you’re going to need to make it big I’m afraid. It’s a collage of 70 images, generated from people moving around Dublin City.

• The top row is Sunday, the next is Monday and so on down until the 7th row, which is of course Saturday.
• Across the image is time of day, with the day being divided into ten parts, or 2.4 hours per column. So the left most column is just after midnight, the centre columns are around midday and the right hand column is from about twenty to ten until midnight.
• Each dot is somebody making a journey on that day of the week, at that time, over the course of the last year.

I can see a few things, and the longer I look at it, the more I see. For instance, the morning commute on weekdays is very clear and it’s very rigid. We go straight to work, and we don’t seem to schedule morning meetings. (fourth column from the left)

Or that people don’t really stay out late on Sunday, Monday or Tuesday, not even until midnight, but they do on Wednesday, Thursday and Friday. (right hand column

Or that people are generally more active on a Saturday morning then a Sunday morning and they stay out a bit later on Saturday night (aka Sunday morning) then on Friday night (aka Saturday morning)

So I guess the question I’m asking is what can you see?

Irish Movement Patterns

This is probably my new favourite thing ever.

Brought to you by the wonderful folks at Openpaths.cc, 46 people who’ve volunteered to be part of the study and the power of R, this is an animation of the data I’ve received to date.

It’s a combination of 70 images, one for each tenth of a day in a typical week. To provide enough points to make it interesting every week is shown at once.

Each dot is a ping of data, generated when the mobile phone of a participant tells openpaths it’s on the move. Each different dot colour is a different individual.

If you like this map, it’s probably worth looking at the presentation I gave at city intersections a few weeks ago too.

The talk will be on the ideas surrounding collaborative consumption and what part that can play in building more sustainable communities and cities.

Look forward to seeing you there!

Happiness Vs Annual Cost of Travel

Things have been a little quiet around here lately, mostly because I’ve been working hard at mastering the art of statistics, specifically the R statistical programming language. Despite the fact that I’m enrolled in a university with all the usual access to courses that entails, it’s simpler and easier for me to just learn the language online via the courses offered by Coursera.

The graph above is just a quick illustration really of something that takes two or three minutes to put together in R. The WHO wellbeing score is across the bottom. Happier people score closer to 25. The Y-Axis is the annual cost of travel for respondents. The width of the boxes shows how many people fall into each Happiness level. Overall the graph really shows that there’s probably no link between how much you spend on travel and how happy you are.

R is fantastic. If you’ve never programmed before, it’s difficult to get into, but it’s free and open source and incredibly powerful. Even just scratching the barest surface, the possibilities it presents are amazing. I’m just getting to the point now where it’s worth playing around with my own data. My favourite things so far is just the ability to clean the data up using a program. Even when the data changes, for instance if I add new points to the openpaths data, I can consistently carry out the same steps and in seconds, everything is ready for me to start using GIS or making graphs. I can have one programme written that deletes all the people with only ten points, and another programme written that gets rid of the points outside Ireland and something else that calculates all my distances. When you have over a hundred thousand data points and excel works at a snails pace, that’s brilliant. I’m not quite there with the regression models and such yet, but I’m getting there. Watch this space.

Coursera itself fascinates me too. University level courses available to everyone. I’m in a university, and they can’t compete with the convenience and standard that coursera is offering me. The courses are tough. Weekly lectures, weekly quizzes, assignments and the pace is very fast. The continuous assessment really forces you to keep up with the course, no saving everything for a last minute cramming session in week 8, if you don’t log in and spend a good six hours a week you won’t get your certificate at the end. On the downside, the online nature of the course forces most testing into the multiple choice format beloved of Americans but which I personally find extremely frustrating, but they are also starting to get there with peer to peer marking of assignments. The collaborative process emerging at a new level with new technology.

It’s been a while admittedly, so happy new year to regular readers!

This is a presentation I’ve put together on all the OpenPaths Data. There is very little explanation I’m afraid, but I don’t think it needs a huge amount. Much of it is updates of maps and graphs I included in previous posts, so if you’re curious take a look at those.

If there’s anything you’re really urgently wondering, leave a message or contact me on twitter/facebook and I’ll answer when I get a moment.

Look at older OpenPaths Posts →

]]> http://researchthecity.com/index.php/openpaths-data-a-presentation/feed/ 1 http://researchthecity.com/index.php/city-intersections-a-speech-by-research-the-city/ http://researchthecity.com/index.php/city-intersections-a-speech-by-research-the-city/#comments Wed, 06 Feb 2013 20:21:33 +0000 Niamh http://researchthecity.com/?p=808

A talk by Research The City

Just a very quick heads up.

On February 12th, I’ll be giving a quick speech at the ninth session of the City Intersections series in “The Twisted Pepper”. It’s on Middle Abbey Street, towards O’Connell Street from the Jervis Luas Stop.

The evening is organised by Dr. Madeleine Lyes as part of her work on urban public discourse. It’s open to everyone, free and it starts at 7pm.

The other speakers will be Rob Kitchin from NUI Maynooth and Mary Mulvihill of Ingenious Ireland. It should be an interesting and thought provoking evening, so I’d love to see you there!

And just for your convenience, I’m including a map below:

View Larger Map

It’s an obvious question, but it’s not one I’ve really looked at in the openpaths.cc data before, so I spent an hour or two today looking at it.

Openpaths doesn’t really distinguish whether you’re moving or not at a particular point in time, what it does is it notices a change in location and records your present position. A”ping” that records your location suggests you’re either travelling or have recently stopped travelling. The more pings in a particular time period, the more likely you were travelling.

And so here is the first of two graphs, the one for Tuesday – Thursday:

Weekday Movement Patterns

Each line represents one person, and the y-axis shows what proportion of “pings” were made at a particular time. For instance the person who is the heavy black line, made 25% of their pings between 6 and 7am. That’s probably their commute to work. Then they travel home, sometime between 4pm and 6pm, possibly stopping for dinner or to go to a gym en route, Or maybe they just work late a lot. The actual movement pattern, showing where they’re going, will give a better idea of that.

The general pattern is pretty clear. Little or no movement from midnight to 6am, then a journey to work, where people stay most of the day, then a less regular journey home.

And the second graph, which shows the same people except on Saturdays and Sundays:

Weekend Movement Patterns

The two heavy lines (the solid black and dark grey dashed) represent the same two people as the represent in the previous graph. You can see how the person with the dashed line clearly has something they always do on weekend mornings. It may be that they work weekends, and rarely take one off, or it may be something like a sports club or religious commitment.

You can also see that the person represented by the solid line does not leave the house for ANYTHING on weekend mornings.

Sensible if you ask me.

See a map drawn using Openpaths Pings →

This is their graph:

Vehicles per Household Vs Households/Acre (Holtzclaw et al. 2002)

Since I have the CSO data to play with, I decided to draw up the same graph for Ireland. As you can see, it’s a messier plot and it has some similarities and some differences. Fewer houses in an area means every house has more cars. The graph is a bit deceptive though, since there are only 18,000 households in areas with densities of greater than 25 houses/acre.

Ireland – Household Density vs Cars per Household

That said, I’m not a huge fan of this graph, so I decided to draw it up again differently. In the graph, which you can see below, I’ve plotted the number of cars per adult against the population density. Each dot represents a single small area. Up to around 1000 adults/square kilometre (4 adults/acre) 4 out of 5 adults will have cars, with most areas having somewhere between 70% and 90% of adults owning a car. Once you start getting population densities above that, you get a much wider range of values – everything from 1 in 5 adults to 9 out of ten adults is reasonably common.

Ireland – Pop Density vs Cars per Adult

I don’t have time to map up this data for another week or two, but I will, and I’ll be interested to see the results. I have a feeling that as usual, you’ll be able to spot the DART and Luas lines, but there’s always a surprise or two lurking around the country.

If you like this post you might also like:

• The Cost of Car Ownership
• Families Living in Cities
• Openpaths Data

My survey was deliberately designed to capture complexity in private vehicle travel. It first asked if you’d travelled in one, two or more than two private vehicles in the last two weeks. Then it asked you about the vehicle you travelled in most. If you said you travelled in two vehicles it asked about the vehicle you travelled in second most, and if you said you travelled in more than two vehicles it asked about all of them at once. Yesterday I put people into the 28 different possible categories based on their answers and their travel patterns. Everything from “travelled in one car only, which they owned” to “travelled in one car and one non-car, owned neither, car travelled in most”.

Most of the time you split people into these sort of categories from a survey and you get 40 – 50% in the largest category, 10- 25% in the next two or three and maybe 1% in the rest of them.

Not the case.

Completely not the case. The highest group was people who travelled in one car only, which they did not own at 20% followed by people who travelled in one car only which they did own at 19%. After that there are another 5 categories in the 5 -20% range. The smaller categories pretty much always involve non-cars, i.e. people travelling in motorcycles and vans.

This is awkward for the model I’m trying to construct at the moment, but it’s excellent both for my research and for the future as a whole. Less than one in four people in the survey travelled only in vehicles they owned, and even those people regularly had passengers or took public transport.

We don’t need to persuade people to share cars, they already do.

Percentage of Time People Spend a Specific Distance from “Home”

I posted the human map of Ireland, today I’m posting two charts drawn using the same data. This time what I’m looking at is how regular peoples patterns are and how far they travel from home.

The answer is that their patterns are very regular.

The chart above shows what percentage of time people spend a specific distance from home. Since I don’t know where people actually live, home is defined in this case as the place they spend most of the their time. As you can see everyone spends somewhere between 25% and 70% of their time at home, with most people clustering around the 50% mark. I suspect if you look into it, people on the lower end have either moved house during the data collection period or regularly spend the night at a girlfriend or boyfriend’s house.

Then most people spend 20% of their time at a second location between 3km and 50km away. That’s their work location. 40 hours a week, works out at 24% of the week, with some allowance for holidays and slight differences in working hours, that’s reasonably safe to assume.

Then you have the holidays cropping up hundreds and thousands of kms away. When you look at the data on an individual level, look at how far people are from home at specific points in the week, it’s also extremely interesting. This is my graph:

Distance From Home by Time of Week

It covers about 8 months worth of data. The line you can see at about 190km is the distance between my house and my parents house. When I get there, I tend not to do very much, so the line is very straight and distinctive. You can also see a few other long distance trips around Ireland, those lines are fuzzier both because they’re once off journeys with fewer dots and because I tend to do more when I go to new places. The diagonal line that happens between 5 and 6 is because I sailed back from Cork once. You can see how much longer it takes the car journeys which are almost straight lines at this scale.

Then you can see the regular weekday pattern of going to work. Since I work in Trinity, and I live fairly nearby this isn’t as distinct as what you can see in other openpathers data. My work trips and my socialising trips can’t really be pulled apart and I don’t start or finish work at the same time each day. I also tend to work from home one day a week. Even so the regularity of the pattern is visible.

Overall, while people do travel around a lot, they tend to go to relatively few specific places.

Some of the earlier posts on the openpaths data are:

• Human Mapping
• Openpaths Data
• Northside Vs Southside, not quite a myth
• Openpaths.cc – Find out where you’ve been