Attribute Parameters

Parameter reference manual for Attribute (field properties).

Parameter List

Required Parameters

Parameter	Type	Description
`name`	`string`	Field name (automatically set when used as key)

Basic Attributes

Parameter	Type	Default	Description
`type`	`string`	`null`	Data type: `int`, `float`, `str`, `date`, `datetime`
`description`	`string`	`null`	Field description text
`logical_type`	`string`	`null`	Logical type annotation (e.g., `email`, `phone`)

Type Attributes (type_attr)

type_attr is a dictionary containing type-related settings:

Parameter	Type	Default	Description
`type_attr.nullable`	`boolean`	`true`	Allow null values
`type_attr.category`	`boolean`	`false`	Whether categorical data
`type_attr.precision`	`integer`	`null`	Numeric precision (decimal places)
`type_attr.format`	`string`	`null`	Datetime format string (e.g., `"%Y-%m-%d"`)
`type_attr.width`	`integer`	`null`	String width (for leading zeros)

ℹ️

Simplified Syntax: Parameters in type_attr can be written directly at attribute level. For example, nullable: false is equivalent to type_attr.nullable: false.

Data Processing

Parameter	Type	Default	Description
`na_values`	`list/string`	`null`	Custom missing value markers (e.g., `"?"`, `["?", "N/A"]`)
`default_value`	`any`	`null`	Default fill value
`cast_errors`	`string`	`"coerce"`	Type conversion error handling: `"raise"`, `"coerce"`, `"ignore"`
`null_strategy`	`string`	`"keep"`	Null value handling strategy: `"keep"`, `"drop"`, `"fill"`

Data Validation

Parameter	Type	Default	Description
`constraints`	`dict`	`null`	Field constraint conditions (`min`, `max`, `pattern`)

Performance & Statistics

Parameter	Type	Default	Description
`enable_optimize_type`	`boolean`	`true`	Enable type optimization (select smallest data type)
`enable_stats`	`boolean`	`true`	Calculate field statistics

System Auto-Generated

These parameters are automatically set by the system, do not set manually:

Parameter	Type	Description
`stats`	`FieldStats`	Field statistics (when `enable_stats=True`)
`is_constant`	`boolean`	Mark fields with all identical values
`created_at`	`datetime`	Creation timestamp
`updated_at`	`datetime`	Update timestamp

Common Examples

Integer Field

age:
  type: int
  nullable: false
  description: "Age"

Categorical Field

gender:
  type: str
  category: true
  description: "Gender"

Float (Specify Precision)

price:
  type: float
  precision: 2
  description: "Price (two decimal places)"

Date Field

birth_date:
  type: date
  format: "%Y-%m-%d"
  description: "Birth date"

Custom Missing Values

workclass:
  type: str
  category: true
  na_values: "?"
  description: "Employment type"

Important Notes

Type Simplification: Use int, float, str, date, datetime (legacy type names auto-converted)
Category Marking: Correctly setting category: true affects data processing and synthesis strategies
Auto-Inference: Unspecified parameters are automatically inferred from data
Performance: For large datasets, disable enable_stats to improve speed

Architecture Data Types