Union

Union module combines given tables into one while keeping unique entry ids. Union is a local module. Like DataIO, this module can be run on the side of Host or Guest, and running this module does not require any interaction with outside parties.

Use

Union currently only supports joining by entry id. For tables of data instances, their header, idx and label column name (if label exists) should match.

When an id appears more than once in the joining tables, user can specify whether to keep the duplicated instances by setting parameter keep_duplicate to True. Otherwise, only the entry from its first appearance will be kept in the final combined table. Note that the order by which tables being fed into Union module depends on the job setting. As shown below:

with FATE-Pipeline:

{
    "union_0": {
            "module": "Union",
            "input": {
                "data": {
                        "data": ["dataio_0.data", "dataio_1.data", "dataio_2.data"]
                }
            },
            "output": {
                "data": ["data"]
            }
        }
    }

with DSL v2:

{
    "union_0": {
            "module": "Union",
            "input": {
                "data": {
                        "data": ["dataio_0.data", "dataio_1.data", "dataio_2.data"]
                }
            },
            "output": {
                "data": ["data"]
            }
        }
    }

Upstream tables will enter Union module in this order: dataio_0.data, dataio_1.data, dataio_2.data .

If an id 42 exists in both dataio_0.data and dataio_1.data, and:

1. ‘keep_duplicate` set to false: the value from dataio_0.data is the one being kept in the final result, its id unchanged.

2. ‘keep_duplicate` set to true: the value from dataio_0.data and the one from dataio_1.data are both kept; the id in dataio_0.data will be transformed to 42_dataio_0, and the id in dataio_1.data to 42_dataio_1.

Here are more [Union job examples] with FATE-Pipeline.

For more example job configuration and dsl setting files, please refer [here].

Param

class UnionParam(need_run=True, allow_missing=False, keep_duplicate=False)

Define the union method for combining multiple dTables and keep entries with the same id

Parameters

• need_run (bool, default True) – Indicate if this module needed to be run

• allow_missing (bool, default False) – Whether allow mismatch between feature length and header length in the result. Note that empty tables will always be skipped regardless of this param setting.

• keep_duplicate (bool, default False) – Whether to keep entries with duplicated keys. If set to True, a new id will be generated for duplicated entry in the format {id}_{table_name}.