Tip:
Highlight text to annotate it
X
There are a number of specific challenges with working with spatial data.
Some of the challenges are the sort of challenges you get with any data.
Say, there are issues of data quality,...
issues of compatibility...
between datasets. But with spatial data I would say that there are...
particular issues, especially with the data I am working with,
around scale...
and also around the accuracy of the data.
So when I'm talking about accuracy I am talking about...
how reliably certain can I be...
that the location of my data, the location point I've got of where something's...
occurring, in this case...
a crime of vandalism, because that's what I'm looking at.
How reliably certain can I be that that is accurate? Has it been...
put in the right spot, or not.
And how reliably certain do I actually need to be?
Because in my research I am particularly interested in differences at a local neighbourhood level,...
therefore,...
which is perhaps one or two streets together,...
or a few streets together.
But I don't need to know precisely what street corners something happens at.
So, for me, I need it to be accurate enough, though I'm pretty certain that it is happening...
in the right neighbourhood, or I can be certain about something about the neighbourhood that it's occurring in,...
but I don't necessarily need to know which house.
And in terms of scale I need to think ... it's a similar question really.
At what level I want to look at my data at.
Now I actually want to look at my data at multiple scales because I don't...
really know at what level the phenomenon is occurring at, and at what level it's having an influence.
When you actually think about what makes up a neighbourhood...
you could think about it being collections of houses, you could think about...
it being collections of streets, you could think about it being...
an area of town that might have a local name.
And each of these things may operate at different scales,...
so you need to then start thinking about well,...
what kind of geographic boundaries can I find that might express those scales.
What's out there that I can use?
And then you need to think about well, what of these boundaries that I can actually get hold of,...
what have they been collected for?
Now I'm working in Scotland, and in Scotland there are pre-available boundaries I can use available through...
the UK Borders service.
And there are things like wards, or what are called...
statistical geographies!
So things like output areas, or data ..... Output areas are the smallest level.
If you are familiar with American data, they are a bit like census blocks...
and they are the smallest level of data at which census data is released...
and they are designed to sort of reflect groups of, I think it's something like between 50...
and 200 people. And so they are the sort of smallest sort of unit of kind of neighbourhood I suppose...
that you can get data at. And they have been produced to be relatively homogenous...
so as to take into account of differences between local neibourhoods, looking at...
socio-demographic variables. So they are a good neibourhood representation.
Ward data on the other hand,...
which tends to be on a bigger scale,...
has been created for the purposes of electoral, electoral purposes so for voting people...
so they tend to vary a lot in size and they tend to change over time,...
whereas output areas stay stable over long time.
So they are a good dataset for me to have a look at because I'm looking at neighbourhoods.
So, in that case, there are specific challenges saying... well are there levels I can get data at.
Is that useful?
The second challenge you run into is that...
if you want to compare your localised data, such as I am using,...
with other data,...
what other data can you get, at what scale? And what you invariably seem to find is...
that different types of socio-demographic data are released at different scales,...
and at different geographical units...
and you need to think about...
what level is that data available at. Is it small enough...
for the problem that you want to look at?
In my case I actually want to look at things at a very small scale...
and actually there's not that much data available at output area level.
A lot of the data I'm interested in is released at data zone level,...
so I've had to look around to see at what level data is available and what data I can use.
And what data might I be able to use that I can perhaps aggregate up at different scales.
And it turns out there is not an awful lot of that but there is data around...
postcodes and addresses that I can look at to get a rough...
approximate how many people have you get in an area.
A further challenge I suppose, coming back to that scale issue,
is what level am I going to want to look at my data at? In my case I'm looking at crime data.
I knew I wanted to look at it at multiple scales, because I was interested in...
what happens at different times, what happens at different scales.
You get different patterns when you look at this data at different scales.
So I couldn't be totally certain starting out...
that if I said, well, just give me...
the postcode for my data,...
that that would be good enough for me to be able to...
aggregate it to different scales. So what I wanted was point level data.
So a specific challenge for me is to get data...
to giving me the exact location of something because once I've got the exact location...
I can then change it or aggregate it up to whatever scale I want.
That gives other challenges, but that is extremely helpful to be able to start at that level.
You wont be able to get all data at that level but if you can,
and in my case I was able to get my ground data, if you can get some level of the exact location...
that can be helpful to you.