if predict_to = grid

• Produces a raster (GeoTIFF) of predicted probability of the outcome. See predict_to parameter for information on defining the resulting resolution/extent.

if predict_to = points

• Produces a csv file of prediction_points with additional columns

• prevalence_prediction: point prevalence prediction (i.e. best guess)
• prevalence_bci_width: 95% prediction interval

If exceedance_threshold is defined, in addition

• exceedance_probability: The probability that true prevalence exceeds exceedance_threshold.
• exceedance_uncertainty: The probability that the predicted class is correct.
• entropy: The entropy (0-1 scale) associated with each prediction. Higher entropy equates to higher uncertainty in classification.

or predict_to = polygons

• Produces the same metrics/fields as if predicting to points. Resulting file is a GeoJSON.

If batch_size is specified, an additional column adaptively_selected with TRUE or FALSE is added, representing whether the location has been selected adaptively.