Thoughts on “Spatial data mining and geographic knowledge discovery – An introduction”

In “Spatial data mining and geographic knowledge discovery – An introduction”, Mennis and Guo articulates the four main methods used in spatial data mining, namely the spatial classification method, spatial association rule mining, spatial clustering and geovisualization, while also explaining the challenges linked with the spatial data mining process.

Although it is true that spatial data mining technologies have greatly evolved over the last few decades, it is always the case that the law is always trailing technological advances, which may allow unethical uses that could compromise the privacy of certain service users, especially from the private sector. While the methods presented in this article seem to be appropriate for many different cases, it could be raised that a partitioning spatial clustering method, which is non-overlapping, might assign a data item to cluster ‘x‘ even though it could have equally been assigned to a cluster ‘y‘, something that could change from one iteration to another.

Interestingly, the conclusion supposes that “the data cannot tell stories unless we formulate appropriate questions to ask and use appropriate methods to solicit the answers from the data”, a notion that could be challenged with the rapid growth of several fields, such as machine learning and artificial intelligence. Although it is hard to conceptualize right now, tt wouldn’t be too far-fetched in the near future where machines could essentially determine by themselves the best algorithms to use in order to classify spatial data from a vast database.

Comments are closed.