Overview
In this step, you select the raster input features DORA will use to identify spatial patterns associated with mineralization.
Input features may include geophysical data, geochemistry grids, geology maps, structural data, or outputs from Data Augmentation modules. These rasters provide the spatial signals the model uses to generate a Prediction Map and calculate the VRIFY Prospectivity Score (VPS).
Note: DORA 2.0 is not yet available. The changes described below will be released in an upcoming update.
What’s Improved
This step now includes automatic raster scoring and feature recommendations.
Instead of manually reviewing correlations or tuning model parameters, DORA evaluates each raster for quality, relevance, resolution, and redundancy. A recommended feature set is generated automatically, helping you move from setup to results quickly.
Advanced configuration has been simplified, reducing the need for manual optimization.
How DORA Evaluates Input Features
Each raster is evaluated using seven data-quality and relevance checks. The final score reflects its overall suitability for the selected AOI and deposit model.
Features are color-coded based on score, making it easy to see which rasters are most strongly recommended:
Green: 81-100%
Yellow: 51-80%
Grey: 0-50%
Recommended input features are automatically selected, allowing you to move forward without manual configuration.
DORA evaluates each input feature for the following:
AOI Coverage
What it measures: The extent to which the raster covers the selected AOI.
How it’s scored: Rasters with more complete coverage receive higher scores, while rasters with large gaps or missing areas receive lower scores.
Model Alignment
What it measures: How well the raster aligns with the type of data used to train the selected deposit model.
How it’s scored: Rasters that are more consistent with the model’s training data receive higher scores.
Mineralization Relationship
What it measures: How strongly raster values are associated with mineralized versus unmineralized learning points.
How it’s scored: Rasters showing stronger statistical relationships with mineralization receive higher scores.
Coarse Resolution Compatibility
What it measures: Whether the raster resolution is too low for the scale of analysis.
How it’s scored: Rasters that are too coarse relative to the AOI or modeling scale receive lower scores.
Fine Resolution Compatibility
What it measures: Whether the raster resolution is excessively high relative to the AOI or modeling scale.
How it’s scored: Rasters that are overly fine and may introduce noise receive lower scores.
Spatial Scale Alignment
What it measures: How well the spatial patterns within the raster match the size and scale of the AOI.
How it’s scored: Rasters with a strong scale match receive higher scores, while scale mismatches reduce the score.
Raster Redundancy
What it measures: The degree of similarity between the raster and other selected rasters.
How it’s scored: Highly correlated rasters receive lower scores to reduce duplication; DORA prioritizes the higher-scoring layer.
How Input Features Are Organized
Input features are organized into six categories for easy review:
Data Augmentation: Rasters generated using VRIFY’s Data Augmentation Modules, which transform raw exploration data into enhanced, continuous input layers for modeling.
Geology: Geology rasters, including those created using the Distance Maps and Fault Disturbance Maps Data Augmentation Modules.
Geochemistry Maps: Geochemical raster layers.
Geophysics: Geophysical raster layers.
Satellite: Satellite-derived raster layers.
Uncategorized: Rasters that do not fit into the categories above.
Understanding Advanced Settings
Prediction Tendency
The Prediction Tendency setting controls how selective the final Prediction Map will be. It determines how closely new areas must resemble your positive learning points to be highlighted.
The default Balanced setting is suitable for most use cases.
You can toggle to:
Conservative:
Prioritizes higher-confidence areas and reduces the number of highlighted targets.
Useful when budgets are limited or drilling decisions must be highly selective.
Aggressive:
Increases coverage by identifying a broader range of potential targets.
Useful in early-stage exploration or when screening large areas.
Resolution
Resolution is automatically determined based on the most common resolution among your selected rasters. This ensures minimal distortion or resampling of your data and maintains consistency across input layers.
In most cases, manual adjustment is not required. Changing the resolution is only recommended when running highly specific or scale-sensitive experiments (e.g., geology-only, high-resolution local modeling).
If you manually adjust the resolution, DORA will re-run the experiment, re-evaluating and re-scoring all input rasters at the new resolution and updating the feature recommendations accordingly.
Step by Step Instructions
Open Select Input Features
From the experiment setup panel, click Step 4: Select Input Features.
Review Recommended Features
DORA automatically displays a pre-selected, recommended feature set.
Features are grouped by category (see categories above).
Next to each feature, you will see:
A recommendation score (0–100).
A corresponding color indicator for quick review.
Hover over the feature on the far right to open Feature Details (AOI, resolution, and other metadata).
To adjust how a raster appears on the map:
Click the eye icon on the left to display the raster.
Adjust Colors and Transparency to highlight key details.
(Optional) Modify Feature Selection
In most cases, we recommend running the experiment with the recommended feature set first.
Click the Recommended Selection slider to turn off automatic selections.
Manually check or uncheck features to create a custom feature set.
(Optional) Change Resolution
This will clear the current input feature selection and rerun the previous step.
Click Apply to save your changes and rerun the previous step.
(Optional) Select Prediction Tendency
All experiments are automatically set to the default Balance setting. Select Conservative or Aggressive from the drop-down, based on your exploration strategy.
Click Apply to save your changes.
Generate Prediction
Click Run Prediction to run the model and create your Prediction Map.
This step takes approximately 30 minutes to complete.
When Should You Modify the Recommended Feature Selection?
In most cases, the automated recommendations are the right starting point.
DORA evaluates each feature using seven data-quality and relevance checks. The final recommendation score reflects the feature’s overall suitability for your selected AOI and deposit model.
Because these suggestions are data-driven, we recommend using them for your initial run.
You might choose to modify the selection if:
You want to test how a specific raster (even with a low score) influences results.
You want to run a targeted experiment using only a specific data category (e.g., geology-only).
Modifying the recommended selection is best used for hypothesis testing or follow-up experiments, rather than for your first run.