Generating a Sample Collection

The next step in the process is to create a Sample Collection based on your data set. A Sample Collection is a set of samples that will either be generated by LACO-Wiki using one of several possible sampling schemes or you can upload your own sample that you have generated outside of LACO-Wiki, e.g., using GIS software.

To start the process in which you generate the sample using LACO-Wiki:

1. Click the button in the Validation Samples section on the Dataset Details screen. This will direct you to the Create a new Sample Collection page (Figure 8).

2. Add the name and a description to your sample collection.

3. Add different sampling methods to your collection.

Depending on type of data that you uploaded (vector or raster), different sampling methods will be available. Currently the following sampling methods are supported:

1. Random Point: This approach creates a sample with a number of points defined by the user that will be randomly distributed over a reference data set, e.g., the data set that you uploaded or a data set that has been shared.

2. Random Pixel: This approach creates a sample with a number of pixels defined by the user that will be randomly selected from a reference data set. Duplicates are not possible.

3. Random Polygon: This approach creates a sample with a number of polygons defined by the user that will be randomly selected from a reference data set. Duplicates are not possible.

4. Polygon at Random Point: This approach creates a sample with a number of polygons defined by the user, which will be randomly selected from a reference data set, where the probability for selection is influenced by the size of the polygons (i.e., large polygons have a higher probability of being selected). Duplicates are not possible.

5. Stratified Random Point: This approach creates a sample with a number of points per class defined by the user, which will be randomly distributed over the specified classes of a reference data set.

6. Stratified Random Pixel: This approach creates a sample with a number of pixels per class defined by the user, which will be randomly selected from the specified classes of a reference data set.

7. Stratified Random Polygon: This approach creates a sample with a number of polygons per class defined by the user, which will be randomly selected from the specified classes of a reference data set.