Hi Guys! That’s the first of articles series where i write about ML.NET framework
Today i write about Data
Data
Data is rappresented by IDataView interface. If you think about Data you can imagine it like a Sql View.
Data is a lazily-evaluated, cursorable, heterogenous, schematized dataset.
Data have two instances:
Schema
A instance of ISchema interface. It contains al the information about the Data view column.
Every column have its name, its type and its metadata.
Use vector<T,N> where T is type and N is size of Data for the rappresentation of the data associated to a row
Cursors
The data view are source of Cursors. Think about Cursors as cursors SQL: a cursors is an object that flowing in the data, one row at time and show data available.
We can have infinite Cursors, because data are immutable.
Nota bene: usually the cursors access a subset of columns, where for efficiency the columns that do not serve the cursor are not calculated.
Data are Lazy!
References: