.rst .pdf

Contents

Data Dictionaries

A data dictionary is the metadata that describes features of a data set. The NumericTableFeature and DataSourceFeature structures describe a particular feature within a dictionary of the associated numeric table and data source respectively. These structures specify:

• Whether the feature is continuous, categorical, or ordinal

• Underlying data types (double, integer, and so on) used to represent feature values

The DataSourceFeature structure also specifies:

• Possible values for a categorical feature

• The feature name

The DataSourceDictionary class is a data dictionary that describes raw data associated with the corresponding data source. The NumericTableDictionary class is a data dictionary that describes in-memory numeric data associated with the corresponding numeric table. Both classes provide generic methods for dictionary manipulation, such as accessing a particular data feature, setting and retrieving the number of features, and adding a new feature. Respective DataSource and NumericTable classes have generic dictionary manipulation methods, such as getDictionary() and setDictionary().

To create a dictionary from the data source context, you can do one of the following:

• Set the doDictionaryFromContext flag in the DataSource constructor.

• Call to the createDictionaryFromContext() method.

Examples

C++:

• https://github.com/oneapi-src/oneDAL/tree/main/examples/daal/cpp/source/datasource/datastructures_aos.cpp

• https://github.com/oneapi-src/oneDAL/tree/main/examples/daal/cpp/source/datasource/datastructures_soa.cpp

• https://github.com/oneapi-src/oneDAL/tree/main/examples/daal/cpp/source/datasource/datastructures_homogen.cpp

previous

Data Sources

next

Data Serialization and Deserialization