Merge pull request #220 from datakind/develop

Release 0.3.4

Author: kaylawilding

CHANGELOG.md

@@ -1,5 +1,9 @@
 # CHANGELOG
 
+## 0.3.4 (2025-06)
+- Update features table (PR #218)
+- Add features table to top shap feature output table (PR #217)
+
 ## 0.3.3 (2025-06)
 - Added unit test for PDP features table (PR 202)
 

README.md

@@ -75,9 +75,10 @@ Note: Since `student_success_tool` is a "library" (in Python packaging parlance)
 ### releases
 
 1. Ensure that all changes (features, bug fixes, etc.) to be included in the release have been merged into the `develop` branch.
-2. Create a new feature branch based off `develop` that includes two release-specific changes:
+2. Create a new feature branch based off `develop` that includes three release-specific changes:
     - bump the `project.version` attribute in the package's `pyproject.toml` file to the desired version; follow [SemVer conventions](https://semver.org)
     - add an entry in `CHANGELOG.md` for the specified version, with a manually-curated summary of the changes included in the release, optionally including call-outs to specific PRs for reference
+    - update the version in the templates and pipelines
 3. Merge the above PR into `develop`, then open a new PR to merge all changes in `develop` into the `main` branch; merge it. Check that main is ahead by 0 from develop. If it is ahead of dev, then merge the main branch back into develop.
 4. Go to the GitHub repo's [Releases](https://github.com/datakind/student-success-tool/releases) page, then click the "draft a new release" button
     - choose a tag; it should be formatted as "v[VERSION]", for example "v0.2.0"

pipelines/institution_id/00-data-assessment-TEMPLATE.py

@@ -25,7 +25,7 @@
 
 # install dependencies, of which most/all should come through our 1st-party SST package
 
-# %pip install "student-success-tool == 0.3.3"
+# %pip install "student-success-tool == 0.3.4"
 
 # COMMAND ----------
 

pipelines/institution_id/01-preprocess-data-TEMPLATE.py

@@ -24,7 +24,7 @@
 
 # install dependencies, of which most/all should come through our 1st-party SST package
 
-# %pip install "student-success-tool == 0.3.3"
+# %pip install "student-success-tool == 0.3.4"
 
 # COMMAND ----------
 

pipelines/institution_id/02-train-model-TEMPLATE.py

@@ -28,7 +28,7 @@
 # we need to manually install a certain version of pandas and scikit-learn in order
 # for our models to load and run properly.
 
-# %pip install "student-success-tool==0.3.3"
+# %pip install "student-success-tool==0.3.4"
 # %pip install "pandas==1.5.3"
 # %pip install "scikit-learn==1.3.0"
 

pipelines/institution_id/03-make-predictions-TEMPLATE.py

@@ -26,7 +26,7 @@
 # we need to manually install a certain version of pandas and scikit-learn in order
 # for our models to load and run properly.
 
-# %pip install "student-success-tool==0.3.3"
+# %pip install "student-success-tool==0.3.4"
 # %pip install "pandas==1.5.3"
 # %pip install "scikit-learn==1.3.0"
 
@@ -46,7 +46,7 @@
 from pyspark.sql.types import FloatType, StringType, StructField, StructType
 
 from student_success_tool import configs, dataio, modeling
-from student_success_tool.modeling import inference
+from student_success_tool.modeling import inference, evaluation
 
 # COMMAND ----------
 
@@ -261,3 +261,25 @@
 dataio.write.to_delta_table(
     result, cfg.datasets.gold.advisor_output.table_path, spark_session=spark
 )
+
+# COMMAND ----------
+
+# MAGIC %md
+# MAGIC ## Log Front-End Tables
+
+# COMMAND ----------
+
+# Log MLFlow confusion matrix & roc table figures in silver schema
+
+with mlflow.start_run() as run:
+    confusion_matrix = evaluation.log_confusion_matrix(
+        institution_id=cfg.institution_id,
+        automl_run_id=cfg.model.run_id,
+    )
+
+    # Log roc curve table for front-end
+    roc_logs = evaluation.log_roc_table(
+        institution_id=cfg.institution_id,
+        automl_run_id=cfg.model.run_id,
+        modeling_dataset_name=cfg.datasets.silver.modeling.table_path,
+    )

pipelines/institution_id/04-register-model-create-card-TEMPLATE.py

@@ -21,7 +21,7 @@
 # we need to manually install a certain version of pandas and scikit-learn in order
 # for our models to load and run properly.
 
-# %pip install "student-success-tool==0.3.3"
+# %pip install "student-success-tool==0.3.4"
 # %pip install "pandas==1.5.3"
 # %pip install "scikit-learn==1.3.0"
 
@@ -109,7 +109,9 @@
 # COMMAND ----------
 
 # Initialize card
-card = PDPModelCard(config=cfg, catalog=catalog, model_name=model_name)
+card = PDPModelCard(
+    config=cfg, catalog=catalog, model_name=model_name, mlflow_client=client
+)
 
 # COMMAND ----------
 

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "student-success-tool"
-version = "0.3.3"
+version = "0.3.4"
 description = "School-agnostic lib for implementing Student Success Tool workflows."
 readme = "README.md"
 requires-python = ">=3.10,<3.13"