Reduce the dimensionality of environmental datasets

Group environmental observations into unlabeled clusters

Assign observations to predefined risk categories (e.g. hazardous vs. safe)

Predict continuous environmental variables such as temperature or pollutant concentration