Skip to content

Commit 47f4140

Browse files
Adding header docs
1 parent 75182a4 commit 47f4140

File tree

1 file changed

+20
-0
lines changed

1 file changed

+20
-0
lines changed

codeflare/pipelines/Datamodel.py

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9,9 +9,29 @@
99
import pickle5 as pickle
1010
import codeflare.pipelines.Exceptions as pe
1111

12+
"""
13+
The core data model structures are defined here. These include the various aspects for creating a DAG, the
14+
input and output to the DAG itself.
15+
16+
The pipeline graph is captured in Pipeline class with the supporting constructs of Node and Edge. There are
17+
different types of nodes, the key ones are EstimatorNode and an AndNode. The details of the kind of nodes
18+
are captured in a separate document that also outlines the type, the firing semantics (when a node gets executed),
19+
and the state of it.
20+
21+
The input and output for the pipeline are captured in the PipelineInput and PipelineOutput classes, which are
22+
supported by the Xy and XYRef classes. The basic data are captured in Xy and XYRef, where XYRef is a holder for
23+
pointers to X and y. The input to the pipeline defines which nodes take what Xy pointers, whereas the pipeline
24+
output defines what nodes produce the outputs.
25+
26+
Finally, the data model allows for morphing of pipeline based on parameterizations, these parameterizations can
27+
be for grid search or for other such similar reasons.
28+
"""
29+
1230

1331
class Xy:
1432
"""
33+
.. _xy
34+
1535
Holder class for Xy, where X is array-like and y is array-like. This is the base
1636
data structure for fully materialized X and y.
1737

0 commit comments

Comments
 (0)