tapas.datasets.data_description.DataDescription

class tapas.datasets.data_description.DataDescription(schema, label=None)

Bases: object

__init__(schema, label=None)

Parameters

schema (list[dict]) – A list of metadata about each column. Each column is represented by a dictionary whose values are the name, type, and on-disk representation of the column.
label (str (optional)) – The name to use to describe this dataset in reports.

Methods

__init__(schema[, label])

param schema: A list of metadata about each column. Each column is represented by a dictionary whose

view(columns)

Returns the same DataDescription restricted to a subset of columns.

Attributes

`columns`	name of all columns, in order.
`encoded_dim`	Number of dimensions the data would have if encoded.
`label`	A label that describes the underlying dataset (and children).
`num_features`	Number of columns.
`one_hot_cols`	List of column names that would be one-hot encoded if encoded.

property columns

name of all columns, in order.

Type: tuple

property encoded_dim

Number of dimensions the data would have if encoded. This assumes ordered and infinite variables will have one dimension, and only finite, unordered variables would be one-hot encoded, where they will require one dimension per category.

Type: int

property label: A label that describes the underlying dataset (and children).

property num_features

Number of columns.

Type: int

property one_hot_cols

List of column names that would be one-hot encoded if encoded.

Type: list[str]

view(columns): Returns the same DataDescription restricted to a subset of columns.