Another ESRI MOOC – Location Advantage

Just started another ESRI online course called the Location Advantage, this one is more directed toward businesses and features exercises which aims to demonstrate how you can use GIS to power business decisions. This is a good opportunity to practice with ArcGIS Online and look at more applied examples in terms of building maps and tools for more commercial business development: The course should take a couple of hours each week for 6 weeks and you can use their ArcGIS Online tool for free.

Banking Exercise

The first exercise is a simple exercise that geocodes a csv list of locations based of address and postal data of a number of banks with some layers of average family size and median income levels.

Some things I’m looking out for with ArcGIS Online is it’s performance, it can be a bit slow to load because of the vector maps. The above example had a full set of shapes for the whole of Canada so when zooming in and out it did take a few seconds for them to load. Another thing I am looking out for is that spatial calculations in ArcGIS can be hit or miss when trying to do some form of spatial processing. As the tool is browser based, it sends calculations operations to be done by a server and so for particularly heavy tasks they can sometimes fail and you lose your progress. This was one of the frustrating issues with the other ESRI course which was based on spatial analysis and the latter exercises became particular repetitive when entering the same settings for the nth time.

ArcGIS Online and ESRI Spatial Analysis MOOC

Just started a Massive Open Online Course provided by ESRI called “Going Places with Spatial Analysis” –

This seemed like a good refresher or intro for spatial analysis as well as an opportunity to pick up using ArcGIS online to build GIS visualisations on web maps.

The first exercise within the course guides the student to create a population map based on some demographics data (click the image below to link to the interactive map!).

Example ArcGIS online map

Example ArcGIS online map

Some noticeable advantages of ArcGIS online is that:

– a cleaner and faster interface compared to ArcGIS desktop

– highly detailed maps instantly available

– easy to share maps

– access to feature layers from ESRI public accounts (allows you to instantly add existing analysis features others have created see below image)

Feature Layers

Feature Layers

Source: Tube feature layer by esriuk.bureau & the population density layer by the UOdocent – Urban Obseratory – Click on map to see the interactive shared map!

This is a really interesting feature of the system to be able to import maps created by others directly and use them to build your own maps with more layers of information helping to inform your own maps and visualisation. The above shows a couple of layers imported, one simply of the tube lines in London on top of a very detailed population density map of London.

Digital Media Analytics – A Reflection

Over the last year I have been working at Manning Gottlieb (a digital media agency) creating interactive reporting gauging the effectiveness of advertisements purchased online. In my first week it was quite an informative time as I started to learnt about how digital advertising worked. Being someone who has been using AdBlockers for many years I had minimal exposure to adverts (primarily to skip through video ads on Youtube videos) so I never gave much thought to what a digital media agency did nor their business models. I learnt it was more than just a company buying some space on popular websites to host a few images and swishy gifs to draw attention to their products/service. Behind each of those adverts they have code embedded to capture information on all the users who have viewed and clicked on those adverts allowing the planners and campaign managers to track how well their advertising are doing. By capturing unique identifiers like an IP address, the servers can tell how many individual people (well unique devices… ) have interacted with those adverts. This gives a solid evidence based data set to look at and measure effectiveness.


Example Data:

Data is extracted via API connections which queried daily reporting metrics from the adservers where a simple report would look like this:



Using Visokio Omniscope I was effectively recreating Excel spreadsheet reports as interactive visualisations.

The process involved:

– downloading the data from source

– transforming the data

– making calculations

– visualising in graphs and tables


This was done in a data driven way with Omniscope providing an all in one environment to do it in.

Further work would be done on integrating other data sources into the report such as bringing client sales data.


Digital Media Report – Click on Image below for link to example report















Opinion: Closure of LFB stations

So just yesterday (on the 9th January 2014), the London Fire Brigade closed ten fire stations in London, so I thought it would be quite apt to write some of my opinions of this after writing a dissertation analysing the incident calls of the LFB. In the dissertation I decided to write quite objectively and in a way more focused on using methods to visualise the incident call out data set and not to probe too much in the political and socioeconomic factors. It was more of a starting point for a “before and after” study which would capture the existing conditions, some relevant contextual information and to build some form of visual analytical tool to  make some form of logical assessment of what it was like before policy had been applied. A lot more could of been done, but I have a habit of optimising my results according of effort needed (in other words I did what I felt was enough and I was holding down a part time job at the same time as writing the dissertation).

My actual personal opinion of the policy to close stations, is in terms of social implications to society as a perceived risk. The driving assurances the government, the London Mayor and the LFEPA is that because incidents are falling we require less firefighters and stations, if it was a simple allocation task I wouldn’t really of thought much about it but when its a spatial allocation task…. well that’s not so simple. This works on one important problem, getting to a fairly random location from set station sites. Statistically speaking if you need to dispatch emergency response from a set of 4 stations in comparison to 3 stations, you will receive a better response from the set of 4 stations, this makes absolute sense to anyone regardless of your level of understanding for spatial complexity, so when you are met with this statement:

“With ten London fire stations set to close this Thursday, 9 January, the Chairman of the London Fire and Emergency Planning Authority (LFEPA), James Cleverly, said today that Londoners will continue to be safe and receive the fastest emergency response in the capital, if not the UK.”

This seems unlikely to me, the only real case I would submit to that being plausible is if the remaining stations each were equipped with fire trucks which were so smart and faster that they could better reach incidents then it might be workable (or better yet squadrons of fire helicopter trucks). Let alone the latter part of the statement, “receive the fastest response in the capital, if not the UK”, well to be quite frank if that is relative to the capital of just London then if you had one station in the London that would still be the fastest response because it is the only response!  As for most parts of the UK, those towns and cities have a lower density of fire station to coverage area and if the capital had 50 station it would probably still have a faster response than other towns and cities in the UK. The truth of this statement holds but in actual consideration of before and after, you will get a slower response than before.

Now, let us analysis the initial reasoning that the fewer incidents means we need fewer stations and pick at this idea for a moment to see if this is understated and actually would work. The easiest way to do so is to think in terms of two monthly scenarios where:

First Month we have 400 call outs addressed by 4 stations.

Second Month we have 300 call outs addressed by 3 stations.

Does it make sense that the average response time in the first month is the same as the second month? It seems on the face of it, in relative ratios that 400 calls to 40 stations = 100 calls per station and 30 calls to 3 stations = 100 calls that well most of it stayed the same so does this mean we have a better or as good a service in this case? The answer though is no, as we do not consider that the size of the spatial area covered by these stations are the same as before. The second month was not covering 75% of the area of the first month, so the response rate is actually going to increase on average. We also ignore the actual network placement of each station in our example here as there isn’t any particular spatial optimisation going on, a true restructuring to best minimise the effect of station closure of a set number of stations will be better served by redeploying station sites to make up for the missing stations as shown below. This will help to optimise how long it takes for a call out to be reached by the emergency response.

Area Comparison

What happens though is that when we close a station without optimising, a quadrant of the first diagram will become an area which will have much weaker response as well as possibly causing greater queues when station resources are the same or less (in reality there are less total fire trucks than before!). These are problems of spatial allocations which actually mean that by decreasing number of sites, even if you maintain the full number of trucks and firefighters as before, the average response will always increase as long as the stations are not operating at full capacity 100% of the time (which they are not).

So….. it won’t be any surprise to those with some level of spatial comprehension if our emergency response time will increase. So what happens then….? Will it be the case that stations be reopened or that new ones be built to replace those when response times increases? Well that will likely be a political problem rather than a spatial one, as was the criteria of the initial proposal for policy change to reduce spending.

My response obviously seems to be against the policy, but not because I feel that is appears flawed but the real problem is that we have a very different situation than the original fire stations network was set up. This spatial allocation task has changed over the years, the road network has grown, congestion levels have changed, population increased and building density has increased. A real forward thinking policy would redevelop and redeploy against the primary issue of how to optimise the task of reaching incidents from a stationary network of sites. This means we really need to improve on the technology being utilised such as more efficient travel systems (SATNAVs and faster speed limits and better trucks/vehicles) or perhaps into more mobile setups like “footloose” fire stations that can be redeployed. Maybe the government could get a logistics delivery company to optimise a brand new station network for them instead? Or better yet… Flying London Fire Brigade drones might just do the trick!

Augmented Reality + Heritage Conservation


An interesting use of augmented reality moving from an initial project we did to augment over a map, we now augmented a physical model. The physical model was created by Billy Dickinson at Central Saint Martins of 19th Century Bloomsbury. To add layers over the physical model we created our application with Unity and Vuforia and incorporated some RAF flyover imagery from the English Heritage Archives from the 1930s and a photorealistic 3D model of 21st Century Bloomsbury, courtesy of Blom UK.


The result is that we have various methods of representations across three centuries to compare against for the same area. I find it somewhat difficult to really explain the idea behind this exhibit based solely on the idea of conservation because it also has a contrast in technology as an aspect. The whole exhibit focuses on using the latest digital technology and using them to simulate or mock up what could of been and how things currently are represented in digital form.


Originally posted on :

We (Kostas, Daniel Lam and I ) have finally finished an augmented reality based app for an exhibit at the “Almost Lost: London’s Buildings Loved and Loathed” exhibition organised by English heritage at Quadriga Gallery, Wellington Arch. The event ( ) is going to be open until 2nd Feb – if anyone is interested in the architectural conservation history of london.

Our work in the exhibit was to build an AR app which adds animations (smoke,people, carriages) to a physical model and also augments it with the current day 3D buildings in the same area and RAF imagery. Screen Shots of the app (in the making) are below. The press coverage of the event can be seen at and .

Credits: 3D model of Bloomsbury by Blom

View original

Visualising the London Fire Brigade Incidents

As a round up to a year of study at UCL studying a master, I had to undertake a dissertation, taking the suggestion of my supervisor and an interest in the Fire Service (as my brother is a fire fighter), I found the openly available data from the LFB quite a helpful resource and based my dissertation on it. It is quite an interesting issue, firstly as a case of using data for driving policy decision and secondly as a topical issue which data has a role of addressing. Data is one of those abstract concepts which is not very well defined but we generally it is a form of evidence of something that has happened in some way or form from our society.

Firstly the data itself is a highly detailed and fairly large just about making it difficult to process with anything less than 8GB RAM. It shows the various call outs that the fire brigade was requested to by calls and documents the type of incidents and the journeys to the incident from the station. This data is comes from two primary LFB data sets being the incident data and the mobilisation records you can find at the London Data store (if you need more recent data you will need to make an FOI request to the LFB, who will subsequently update those sources). The data is uploaded to a postgresql database for storage and ease to query the data for visualisation purpose where any sizeable data driven visualisation really needs a robust backend storing large data sets and of various types of data.


Here created in processing is the busiest day for 2012 sped up:


I particular chose this day just to emphasis a maximum capacity which highlights the most stretched capacity the service will receive, this to me is a particularly important point as deriving the capacity from a mean use will not be good in the social context as what is the point of having a response service that cannot due with disaster level incidents and maintain excellent service? An interesting note is that the majority of calls were to flood incidents, something not primary to fire response which somewhat shows a change in the usage of our services.

As the primary output, I decided to use Omniscope to visualise and deploy the output as it was able to combine the traditionally GIS output with interactivity, as well as facilitate the use the extract, transform and load function of the software to process. I am particularly fond of approaches to data visualisation which actually maintains the level of depth and access to raw data and found this can communicate detail which may be missing.

The research does not really aim to tell whether or not the policy is good or bad, or even if we need more or less fire response. It is supposed really create an environment to cast a critical view and one that is objective. Hence the aspect of producing a interactive detailed visualisation and a back end database behind it being quite handy.


Here’s a poster I made for the UCL Bartlett Postgraduate MRes conference (very briefly going over it):


Future of Visualisations

After completing projects and nearing the end of my year of study of analysing and visualising data, there has been quite a change in the world of data and the internet. Just a year ago I remember seeing the first Facebook network graph but now there is a story every week of some amazing visual spatial graphic, the most recent being this creation from WeAreData from Betc & Ubisoft:

This concept of a hyper-connected world, big data, real time APIs, analytics and smart cities are fascinating areas which blur the distinction between the real world and that of the digital world, something explored through technologies of mobile phones, tablets and augmented reality. Future release of the Google Glass will help to drive this idea of going forward and the methods of visualising this cyber world will become increasingly common. Google themselves aren’t just looking at nifty bits of hardware but also their Google maps are receiving a revamp with more API options to make visualising spatial data more easier and more visually appealing as well.

My course mates and I have been able to explore a range of these and have created projects using augmented reality + Unity, javascript in Google Maps and Processing. All with the sole purpose of visualising spatial data in very different ways, whilst I have learnt so much, this has opened up a whole new world of even more things to learn.

UCL IRIS data & Google Maps:

UCL IRIS data & Processing:

Augmented Reality and Unity Map:

What is most interesting I have found is there huge gap in what is being used in an analytical sense and what is just for visual appeal. Some of the most intriguing examples such as WeAreData is developed by computer game developers solely on an idea from their upcoming release of Watchdogs ( a game about a computer hacker). It may not be the most academic or professional viewpoint to take but  the best examples are from computer games pioneering some of the most unique visual maps out there, here’s another upcoming title Tom Clancy’s Division which features concepts of augmented reality, hovering drones and technologies which are partially available in the real world. There really is something there that computer games have to offer which seems to be ahead of every other area of mapping, data visualisation and analytics which professional bodies and businesses just aren’t able to do and could be an avenue for analytical methods to change drastically. Perhaps we all will need to start hiring game developers to make that impact in communicating the data and information.

_characterise _space

a blog about spatial analysis and software development

City Remade

Understanding spatial data trends

Matthew Hopkinson - The Local Data Company

Director of Local Data Company. Tracking and reporting on the UK's 'High Street'. News, reports, tips, facts and commentary on what is happening in UK retail.


Understanding spatial data trends

laboratory of fun

the co-joined efforts courses in the college of architecture at iit

Ripped-off Britons

Understanding spatial data trends

Luke Avery Portfolio

Hungry Hungry Developer

Econometrics Beat: Dave Giles' Blog

Understanding spatial data trends

Ferrara Fabio

Game & Application Developer, 3D Animator, Composer.

Mapping Complex Information. Theory and Practice

From an Information Design Perspective


Understanding spatial data trends

MAPCITE Location Intelligence Blog

Understanding spatial data trends

Human Plus Design

Helping technology understand humans.

Sketchpad Blog

Understanding spatial data trends


Data & Art Miscellanea from Jer Thorp

Stuart Hayes (mongopi)

Stuart Hayes - dabbling with MongoDB, Raspberry Pi & general bigdata stuff

Adam Dennett

Understanding spatial data trends


Understanding spatial data trends

Northern Insight

Thoughts and experiences about working life

Simon Rogers

Data journalism and other curiosities

Claire Bailey Ross

exploring museums, digital technology and user experiences

Data Animator

on a mission to humanise data

Stand-Up Economist

Understanding spatial data trends

Cyber Geography

Understanding spatial data trends

Ordnance Survey Blog

Understanding spatial data trends

Reviving Economics

Understanding spatial data trends

Understanding spatial data trends

3rd platform

Understanding spatial data trends

Understanding spatial data trends


Understanding spatial data trends


R news and tutorials contributed by (573) R bloggers

Understanding spatial data trends

City of Blinding Lights

Research blog on all things cities, data, mapping, urban issues and human rights

Just another site

Mapping London

Understanding spatial data trends

Understanding spatial data trends

Sociable Physics

Because "social physicist" is not an oxymoron.

Understanding spatial data trends

digital urban

Understanding spatial data trends

Understanding spatial data trends

A Science of Cities

Understanding spatial data trends

Data Pointed

Understanding spatial data trends

Greg Mankiw's Blog

Understanding spatial data trends

Interesting data

Understanding spatial data trends

evolving Urban

urban analysis | urban modelling | pedestrian simulations | data visualisation

The CASA Blog Network

Understanding spatial data trends

S. Bala Murugan's Blog

Fortunae Meae...!


Get every new post delivered to your Inbox.

Join 78 other followers

%d bloggers like this: