CategoricalDataSet Class

A dataset is composed by a set of categorical variables, whose list is returned by property Variables, and a matrix, returned by property Data, consisting of the data observed for such variables at a given collection of individuals. Each matrix column is associated to one of the categorical variables under study, while the rows of the matrix are associated to the individuals.

Instantiation

New instances of the CategoricalDataSet class can be initialized from previously encoded data, through method FromEncodedData(ListCategoricalVariable, DoubleMatrix), or by encoding a data source, see, for example, Encode(TextReader, Char, IndexCollection, Boolean, DictionaryInt32, Categorizer, IFormatProvider).

The source can contain information about categorical or numerical variables observed at a given instance. Encoding methods take into account numerical variables by delegating their discretization to special categorizers. If needed, categorizers can be identified by splitting the range of the numerical data into multiple intervals in order to minimize the intra-interval heterogeneity of the given target, see, for example, CategorizeByEntropyMinimization(TextReader, Char, IndexCollection, Boolean, Int32, IFormatProvider)

Reverting an encoding operation is provided by method Decode.

Parts of the CategoricalDataSet can be selected through indexers, see, for example, ItemIndexCollection, IndexCollection.

Disjunctive forms

Data about a categorical variable can be represented in disjunctive form by splitting the information for the variable in as many binary variables as the number of variable categories. The disjunctive representation of a CategoricalDataSet is returned by Disjoin. Supplementary data, i.e. data containing information about the same variables observed at different individuals, can be obtained by method Disjoin(DoubleMatrix).

Serialization

Categorical data sets can be represented as JSON strings, see JsonSerialization.

Reference