131Algorithmic Design Considerations for Geospatial and/or Temporal Big Data
When can we use ltering effectively under which scenarios and which math-
ematical and statistical assumptions? How much ltering is allowed before the solu-
tion completely degrades and becomes nonrepresentative of the actual solution? Can
we fully describe the relationship between the amount of ltering being applied, the
amount of data, and the quality of the solution? What types of ltering are allowed
under which types of data?
It is well understood that aggregation is not only an effective form of data
reduction but also may be applied to any data leading to very wrong results. For
instance, making an assumption about the mode of the data and applying a given ...