[R-sig-Geo] vulnerability mapping

Juta Kawalerowicz juta.kawalerowicz at nuffield.ox.ac.uk
Sun Oct 23 23:53:26 CEST 2016


Thanks Tina and Jacek for your messages!

More about the problem (which hopefully I can illustrate this better now).
I want to show that the area where 2011 rioting in London started was
particularly good for a quick spread of the rioting given that it was on
the intersection of high density of shopping malls which are targeted
during rioting and socio-economic deprivation. So no modelling of riots per
se just showing that the area where the (initially peaceful) protest which
later turned violent happened fell into a vulnerable zone.

What data I have:

spatial points showing malls locations (see here
<https://www.dropbox.com/s/9e9dnjcy0182txf/picture1.tiff?dl=0>)
spatialpolygon dataframe which has a column with indices of deprivation for
each polygon (see here
<https://www.dropbox.com/s/3yc3sqyb9ltj7ir/picture6.tiff?dl=0> shows
quintiles)

The concept so far is to:

1. Create hotspots from shopping malls locations (here
<https://www.dropbox.com/s/tl8x9xabxcyzp1v/picture2.tiff?dl=0>)
2. Convert hotspots into spatial polygons (here
<https://www.dropbox.com/s/pd7i0lz90p5pbbx/picture3.tiff?dl=0>)
3. Calculate the mean value of deprivation within each hotspot polygon.
Before I would need to create a raster out of deprivation in neighbourhoods
(here <https://www.dropbox.com/s/m43t38wywllmh2w/picture4.tiff?dl=0>)
4. The final result would be to create a column based on multiplication of
shopping map hotspots and mean value of deprivation in that hotspot polygon
(here <https://www.dropbox.com/s/qune4w812sikwmr/picture5.tiff?dl=0>)

I realise that this can seem quite arbitrary as a method so I was wondering
whether someone would now a more standard procedure or R packages which are
normally used for such tasks?

Juta

On Fri, Oct 21, 2016 at 2:59 PM, Tina Cormier <tcorms at gmail.com> wrote:

> Hi Juta,
>
> It seems to me that what you have are potential predictors of riots. I
> think you'll also need some riot location data - that is, if you are trying
> to model vulnerability (as Jacek said) to identify important variables. If
> you are just trying to create a map that shows potential areas (without
> modeling - as in, you already KNOW that these are good predictors), then I
> would probably do something simple, like create a ranking system for
> socioeconomic status and distance from malls - add them together to ID hot
> spots. Without knowing your goal, it's probably not worth going into too
> much detail on those methods, but I would start by interpolating a raster
> that shows distance to malls (ranked), and combine with census rankings. It
> seems to me these two predictors alone are not sufficient, but if that's
> what you're going with, this would be my simple approach.
>
> Good luck!
> Tina
>
> On Fri, Oct 21, 2016 at 8:41 AM, Jacek Stefaniak <
> jacek.stefaniak at gmail.com> wrote:
>
>> Hi,
>>
>> Honestly speaking I do not have clear idea what you will plan to do. You
>> want to just create map with highlighted areas showing hypothetical
>> "better
>> conditions" for riots, or you want to use a model for finding such areas
>> and explore potential variables causing higher risk? In any of this two
>> cases, you will need slightly different data and approach... Can you care
>> to elaborate?
>>
>> PS: Możesz napisać do mnie po polsku na prywatny adres, jeśli tak Ci jest
>> łatwiej ;)
>>
>>         [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-Geo mailing list
>> R-sig-Geo at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>
>
>

	[[alternative HTML version deleted]]



More information about the R-sig-Geo mailing list