Dictionary
Last updated
Last updated
In ProcessMIX, Dictionary is a tabular data set (a spreadsheet) used for storing lookup information, which you can create directly in your project for easy access and maintenance. Columns in a ProcessMIX dictionary have data types, so that they can be used for structured information storage and retrieval - an essential requirement in automated data processing.
When a dictionary is created, a corresponding backing is also automatically defined in the same project, which contains a separate property for each table column,. This way dictionary entries can be accessed programmatically from flows in a standard fashion, as any other processed data.
When you edit dictionary parameter names and values, the backing data structure is automatically updated as well, in order to keep it in sync with the dictionary settings.
There are two types of dictionaries in ProcessMIX:
Custom dictionary—a general-purpose lookup table, which you can use for information retrieval by adding a corresponding node in a flow.
Reason code dictionary—a specialized lookup list for providing explanatory reason codes and descriptions when calculating predictive scores with the help of assets.
The type is specified via a configuration property in the dictionary builder.
Custom dictionaries can have any number of columns, but they must be of simple data types: String, Integer, Float, Boolean, or Date.
For example, you could define a custom dictionary to store a list of country codes, names and their currency codes:
Then you can add a reference to it as a node to your flow and be able to retrieve a full or filtered list of country records to use in the rest of the flow as you wish.
This dictionary type is used as a supporting tool when developing scorecards. It helps to produce a human- and machine-readable explanation of the reasons why a particular score was calculated by the model.
It is restricted in a sense that it needs have few standard columns, which are automatically added as properties to the auto-generated backing data structure:
Rank (Integer)
Name (String)
Code (Integer)
Message (String)
Additional custom properties/columns can also be added, but the above mentioned ones are required for the proper scorecard implementation.
You can learn more about creating and editing dictionaries in the section.