Add a 'categorical_values' option to the config that will create alternate versions of the specified columns that have evenly-spaced values instead of the categorical values in the original column. For each property that represents a categorical value, this option must include all of the possible values and provide an order that the values should be sorted into.
For example, the permafrost and ground ice dataset includes a categorical EXTENT property that can have one of four values, represented by their first letter: Continuous, Discontinuous, Sporadic, Isolated patches. Adding the following 'categorical_values' option to the config will result in a new column (named staged_normalized_EXTENT) mapping those values to 4, 3, 2, and 1 respectively:
So long as the list of all possible values for the column is known and specified, this should work for categorical properties of any data type, so it resolves #48.
Add a 'categorical_values' option to the config that will create alternate versions of the specified columns that have evenly-spaced values instead of the categorical values in the original column. For each property that represents a categorical value, this option must include all of the possible values and provide an order that the values should be sorted into.
For example, the permafrost and ground ice dataset includes a categorical EXTENT property that can have one of four values, represented by their first letter: Continuous, Discontinuous, Sporadic, Isolated patches. Adding the following 'categorical_values' option to the config will result in a new column (named staged_normalized_EXTENT) mapping those values to 4, 3, 2, and 1 respectively:
So long as the list of all possible values for the column is known and specified, this should work for categorical properties of any data type, so it resolves #48.