Currently we are using the UK Census Output Areas (OAs) as unit throughout the project, iE the modelling and the outputs are given a this level of aggregation.
This came for obvious reasons related the availability of some of the data, but also brought to limitations and approximations on different levels (IE for example needing to interpolate data at a different resolution, or obtaining results which don't reflect the actual geography/aggregation on the ground - Lands don't vote).
We suggest to consider different spatial aggregation levels, like on a hexagonal grid (I have stumbled on this recently, for example) or other.
Please @darribas and @martinfleis feel free to add pros and cons in comments. @yongrenjie's input will be really appreciated too!
Currently we are using the UK Census Output Areas (OAs) as unit throughout the project, iE the modelling and the outputs are given a this level of aggregation. This came for obvious reasons related the availability of some of the data, but also brought to limitations and approximations on different levels (IE for example needing to interpolate data at a different resolution, or obtaining results which don't reflect the actual geography/aggregation on the ground - Lands don't vote). We suggest to consider different spatial aggregation levels, like on a hexagonal grid (I have stumbled on this recently, for example) or other.
Please @darribas and @martinfleis feel free to add pros and cons in comments. @yongrenjie's input will be really appreciated too!