Spatial and Advanced Analytics
What to know
Spatial tools create and manipulate geometry: Create Points (from lat/long), Spatial Match (point-in-polygon, distance), Spatial Info (area, centroid, distance), Buffer, Trade Area, and spatial object handling. Advanced analytics include the predictive tools (Linear Regression, Logistic Regression, Decision Tree, Forest Model, Score), clustering (K-Centroids, K-Nearest Neighbors), and time-series/forecasting. Understand the distinction between model-building tools (with an O anchor) and scoring tools (with an S/input anchor), and how a model object flows between them.
Traps
- Coordinate order: Create Points takes X (longitude) and Y (latitude); swapping them places points in the wrong hemisphere.
- Spatial reference systems: Mixing projections can misalign matches; ensure consistent spatial reference.
- Model vs. Score anchors: Building a model outputs a model object; scoring consumes it plus new data.
- Target field type: Logistic Regression requires a categorical/binary target; Linear Regression requires numeric.
- Spatial Match types: Point-in-polygon, polygon-polygon, and distance-based matches behave differently; read the specified method.
Rule of thumb
For any predictive workflow, separate "train" from "score": one branch builds the model, the model object feeds the Score tool alongside new records. Keep field sets identical between training and scoring.
Worked example
To count customers inside sales territories: Create Points from customer Latitude/Longitude (X = longitude, Y = latitude), then Spatial Match customers (points) to territories (polygons) using point-in-polygon, then Summarize Count grouped by territory.