Overview
In this step, you’ll define and visualize your final exploration targets using the VRIFY Prospectivity Score (VPS) generated by your predictive model. You'll adjust threshold and clustering settings to group high-prospectivity zones into distinct targets, then evaluate them using your geological knowledge and project context.
The Identify Targets step is where DORA begins to open the black box, showing how the AI’s predictions connect back to your actual data. Using Feature Importance plots (SHAP values), DORA reveals which input layers are influencing each target, helping geoscientists evaluate predictions through a geological lens.
Understanding Target Identification
This step takes the continuous output from your model (prospectivity scores) and converts it into clearly defined targets. By adjusting thresholds and clustering parameters, you influence which zones are identified as targets and how they’re grouped.
While the clustering method may resemble what was used during model training, it’s important to note the difference: here, clustering is based on the VRIFY Prospectivity Scores, not the original learning points. DORA filters the AOI by your selected VPS threshold, then groups spatially connected high-prospectivity areas into target clusters.
This accounts for the fact that different regions within your AOI may be influenced by different geological drivers (e.g., geophysics in one area, faults in another), and allows for more meaningful target delineation.
Why This Step Matters
Geoscience Perspective
Target clustering enables you to group predictions into geologically meaningful exploration zones. Since different areas may be influenced by different mineralization controls, this step allows you to apply domain knowledge and evaluate whether the generated targets align with known geological patterns.
AI Perspective
This step turns your model results into clear exploration targets. By adjusting threshold and clustering settings, you control how targets are grouped, filtering out noise and preventing the output from breaking into too many small, disconnected zones.
Step by Step Instructions
Open Step 6: Identify Targets
After completing the predictive model, this step will display VPS results and allow you to generate exploration targets from the output.
Review Prospectivity Score Visualization
The view panel displays a color gradient representing the VRIFY Prospectivity Score (VPS):
Evaluate Results
Use your geological knowledge to assess whether high-prospectivity zones align with known or expected mineralization features in your project. Result graphs appear on the right of the screen to assist with interpretation.
Adjust Target Identification Parameters
These parameters influence how prediction scores are grouped into discrete exploration targets:
VPS Threshold: Higher thresholds focus on high-certainty targets. Lower thresholds offer broader views but may include more noise.
Minimum Group Size: Controls spatial separation between target points
Minimum Number of Points in Group: Filters out very small clusters.
Use the table below to guide your parameter settings based on your geological context:
Generate and Review Targets
Once your parameters are set, targets will appear in the view panel, each color-coded as a distinct cluster. Adjust the parameters above as needed to refine them:
Generate Feature Importance Labels
Click Generate Feature Importance Labels to assign one label per target cluster. This will help with later interpretation and analysis.
Review and Export Results
Review the output graphs in the Results panel:
Tips & Considerations
VPS Threshold
Adjusting the VPS slider at the bottom of the screen does not affect Feature Importance (SHAP) clusters:
However, adjusting thresholds will update the Feature Importance (SHAP) outputs accordingly:
Iterate as Needed
You can adjust parameters and regenerate targets multiple times before finalizing your Prediction Map.
Compare Experiments
If you have completed two experiments that use the same asset, you can compare them on a single Prediction Map. Learn more in Compare Experiments.
Learn More
Previous Steps:
Next Step:
Interpret DORA’s Result Graphs:
Still Have Questions?
Reach out to your dedicated DORA contact or email support@VRIFY.com for more information.