Decide fate of -viz starters and viz tool

[astrojuanlu]

Hi @ankatiyar and @DimedS , I am not sure if it is worth to keep the viz starters after removing ET. The difference I see are the plotly charts. I feel we can keep the viz starters and rename them to spaceflights-pandas and spaceflights-pyspark to get the plots. What do you think ?

Originally posted by @ravi-kumar-pilla in #262 (comment)

About keeping "Viz" as a tool in kedro new, initially the idea was to add make it add kedro-viz as a dependency, but... it was already a dependency of the non -viz starters!

kedro-starters/spaceflights-pandas/{{ cookiecutter.repo_name }}/requirements.txt

Line 6 in adc6eb8

kedro-viz>=6.7.0

Therefore we need to

• Agree on having only spaceflights-pandas and spaceflights-pyspark, optionally merging the contents of -viz
• Agree on what to do with the viz tool in kedro new

[astrojuanlu]

Difference between picking the Viz tool no/yes, including 1-5 and example pipeline in both cases:

--- tree_no_viz.txt	2025-02-13 10:36:20
+++ tree_yes_viz.txt	2025-02-13 10:36:28
@@ -1,4 +1,4 @@
-kedro-no-viz/
+kedro-yes-viz/
 ├── README.md
 ├── conf
 │   ├── README.md
@@ -6,7 +6,8 @@
 │   │   ├── catalog.yml
 │   │   ├── parameters.yml
 │   │   ├── parameters_data_processing.yml
-│   │   └── parameters_data_science.yml
+│   │   ├── parameters_data_science.yml
+│   │   └── parameters_reporting.yml
 │   ├── local
 │   │   └── credentials.yml
 │   └── logging.yml
@@ -30,7 +31,7 @@
 ├── pyproject.toml
 ├── requirements.txt
 ├── src
-│   └── kedro_no_viz
+│   └── kedro_yes_viz
 │       ├── __init__.py
 │       ├── __main__.py
 │       ├── pipeline_registry.py
@@ -40,7 +41,11 @@
 │       │   │   ├── __init__.py
 │       │   │   ├── nodes.py
 │       │   │   └── pipeline.py
-│       │   └── data_science
+│       │   ├── data_science
+│       │   │   ├── __init__.py
+│       │   │   ├── nodes.py
+│       │   │   └── pipeline.py
+│       │   └── reporting
 │       │       ├── __init__.py
 │       │       ├── nodes.py
 │       │       └── pipeline.py
@@ -53,4 +58,4 @@
     │       └── test_pipeline.py
     └── test_run.py
 
-24 directories, 30 files
+25 directories, 34 files

Relevant content differences between the two:

diff --color=auto -u kedro-no-viz/pyproject.toml kedro-yes-viz/pyproject.toml
--- kedro-no-viz/pyproject.toml	2025-02-13 10:35:55
+++ kedro-yes-viz/pyproject.toml	2025-02-13 10:36:09
@@ -4,7 +4,7 @@
 
 [project]
 requires-python = ">=3.9"
-name = "kedro_no_viz"
+name = "kedro_yes_viz"
 readme = "README.md"
 dynamic = ["version"]
 dependencies = [
@@ -12,13 +12,14 @@
     "jupyterlab>=3.0",
     "notebook",
     "kedro[jupyter]~=0.19.11",
-    "kedro-datasets[pandas-csvdataset, pandas-exceldataset, pandas-parquetdataset]>=3.0",
+    "kedro-datasets[pandas-csvdataset, pandas-exceldataset, pandas-parquetdataset, plotly-plotlydataset, plotly-jsondataset, matplotlib-matplotlibwriter]>=3.0",
     "kedro-viz>=6.7.0",
-    "scikit-learn~=1.5.1"
+    "scikit-learn~=1.5.1",
+    "seaborn~=0.12.1"
 ]
 
 [project.scripts]
-"kedro-no-viz" = "kedro_no_viz.__main__:main"
+"kedro-yes-viz" = "kedro_yes_viz.__main__:main"
 
 [project.entry-points."kedro.hooks"]
diff --color=auto -u kedro-no-viz/requirements.txt kedro-yes-viz/requirements.txt
--- kedro-no-viz/requirements.txt	2025-02-13 10:35:56
+++ kedro-yes-viz/requirements.txt	2025-02-13 10:36:09
@@ -1,7 +1,8 @@
 ipython>=8.10
 jupyterlab>=3.0
-kedro-datasets[pandas-csvdataset, pandas-exceldataset, pandas-parquetdataset]>=3.0
+kedro-datasets[pandas-csvdataset, pandas-exceldataset, pandas-parquetdataset, plotly-plotlydataset, plotly-jsondataset, matplotlib-matplotlibwriter]>=3.0
 kedro-viz>=6.7.0
 kedro[jupyter]~=0.19.11
 notebook
 scikit-learn~=1.5.1
+seaborn~=0.12.1

So given that we were installing Kedro-Viz in both cases anyway and that the content differences are quite minor, I contend that

• Kedro-Viz should continue to be installed in all cases,
• we should add information on the README of the starters on how to launch Kedro-Viz ("run kedro viz run and you will see [screenshot] in the browser")
• while we're at it, we should mention the VS Code extension too,
• we should add the Plotly and matplotlib datasets as well as seaborn if the user asks for an example pipeline,
• and the Viz tool should be removed (given that it will be installed anyway)

If I understand correctly, this means merging spaceflights-pandas-viz with spaceflights-pandas, and spaceflights-pyspark-viz with spaceflights-pyspark.

Thoughts?