Multi-class Classification examples by mshtelma · Pull Request #28 · mlflow/recipes-examples

mshtelma · 2022-12-19T15:27:58Z

No description provided.

Signed-off-by: mshtelma <mshtelma@gmail.com>

…dded hyperparameter tuning to the multi-class example

sunishsheth2009 · 2022-12-19T17:46:44Z

requirements.txt

@@ -1,4 +1,4 @@
-mlflow>=2.0


Is there a reason we removed the mlflow requirement?

Thanks, good catch! Added it back.

sunishsheth2009 · 2022-12-19T17:47:42Z

requirements.txt

+scikit-learn==1.2.0
 ipykernel>=6.12
 ipython>=7.32
+flaml


Do we need flaml overall? Or do you think it is a good idea to add it in the notebook or create a sub requirement for this? Thoughts?

That's a good point. I think there are other examples that use FLAML by default (at least I saw some PRs), so this might be a good idea. This is not a big dependency. I am happy to move it inside the multi-class folder as well.

I recall flaml is problematic to install with lightgbm installation difficult. Has this been fixed? If not, prefer to not including it. It's not used in this example anyway.

sunishsheth2009 · 2022-12-20T00:10:31Z

multi-class-classification/README.md

@@ -0,0 +1,26 @@
+# Binary classification: Is this bottle of wine red or white?
+This is the root directory for an example project for the
+[MLflow Classification Recipe](https://mlflow.org/docs/latest/recipes.html#classification-recipe).


[MLflow Classification Recipe] --> [MLflow Multi Class Classification Recipe]

What do you think?

sunishsheth2009 · 2022-12-20T00:11:23Z

multi-class-classification/notebooks/databricks.py

+# COMMAND ----------
+
+# MAGIC %pip install -r ../../requirements.txt
+# MAGIC %pip install git+https://github.com/mshtelma/mlflow.git@multiclassclassification


Remove this once MLFlow is released.

We just released MLflow 2.1.0 so we should be good to remove this now :)

jinzhang21 · 2022-12-21T23:42:11Z

multi-class-classification/notebooks/databricks.py

+# COMMAND ----------
+
+r.run("split")


Can we do a bit of EDA here before proceeding to split?

jinzhang21 · 2022-12-21T23:42:57Z

multi-class-classification/notebooks/databricks.py

+
+# COMMAND ----------
+
+trained_model = r.get_artifact("model")


Can we use the model to make a prediction on one hand-generated input example here?

jinzhang21 · 2022-12-21T23:43:39Z

multi-class-classification/profiles/databricks.yaml

+  # For different options please read: https://github.com/mlflow/recipes-classification-template#ingest-step
+  using: csv
+  loader_method: load_file_as_dataframe
+  location: "./data/iris.csv"


Can this be an open access Databricks Delta table? Does it exist?

jinzhang21 · 2022-12-21T23:45:47Z

multi-class-classification/profiles/local.yaml

@@ -0,0 +1,31 @@
+experiment:
+  name: "sklearn_classification_experiment"


Update the experiment name to not collide with binary classification example

jinzhang21 · 2022-12-21T23:46:28Z

multi-class-classification/profiles/local.yaml

+INGEST_SCORING_CONFIG:
+  # For different options please read: https://github.com/mlflow/recipes-classification-template#batch-scoring
+  using: csv
+  location: "./data/iris.csv"


It'd be nice to use a different dataset for scoring, if such dataset exists. Or you can create one. See https://github.com/mlflow/recipes-examples/blob/main/classification/profiles/local.yaml#L26 for example.

jinzhang21 · 2022-12-21T23:51:26Z

multi-class-classification/steps/ingest.py

+        import pandas
+
+        df = pandas.read_csv(file_path, sep=",")
+        df["class"] = df["class"].astype("category").cat.codes


What does this line do? What happens if there's no such conversion?

jinzhang21 · 2022-12-21T23:52:23Z

multi-class-classification/steps/transform.py

+    transformers.
+    """
+
+    return None


Add a comment indicating this will result in an identical transformer.

jinzhang21 · 2022-12-21T23:52:45Z

multi-class-classification/steps/split.py

+    :return: A Series indicating whether each row should be filtered
+    """
+
+    return Series(True, index=dataset.index)


Add a comment to indicate this doesn't filter out anything.

jinzhang21 · 2022-12-21T23:53:34Z

multi-class-classification/tests/transform_test.py

@@ -0,0 +1,10 @@
+from steps.transform import transformer_fn


Why introduce an empty file multi-class-classification/tests/train_test.py above?

jinzhang21 · 2022-12-21T23:55:10Z

requirements.txt

+scikit-learn==1.2.0
 ipykernel>=6.12
 ipython>=7.32
+flaml


I recall flaml is problematic to install with lightgbm installation difficult. Has this been fixed? If not, prefer to not including it. It's not used in this example anyway.

mshtelma and others added 7 commits December 5, 2022 22:49

Multiclass classification example using iris dataset

035d432

Signed-off-by: mshtelma <mshtelma@gmail.com>

Text Classification

78f591d

Text Classification Example

f2e48f1

Small fixes

433c7ce

Initial Commit

d621a40

Added NLP examples (TF-IDF and HuggingFace sentence embeddings) and a…

66d95f7

…dded hyperparameter tuning to the multi-class example

Renamed classification to binary-classification

a447f20

sunishsheth2009 reviewed Dec 19, 2022

View reviewed changes

mshtelma added 2 commits December 19, 2022 18:59

Removed NLP examples.

95a2a2b

Removed NLP examples.

fa00772

mshtelma changed the title ~~Multi-class Classification and Text Classification (TF-IDF and Transformers) examples~~ Multi-class Classification examples Dec 19, 2022

Improved formatting

8db1e3e

sunishsheth2009 reviewed Dec 20, 2022

View reviewed changes

sunishsheth2009 requested a review from jinzhang21 December 21, 2022 18:15

jinzhang21 reviewed Dec 21, 2022

View reviewed changes

		@@ -0,0 +1,31 @@
		experiment:
		name: "sklearn_classification_experiment"

Conversation

mshtelma commented Dec 19, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants