Merge pull request #143 from sassoftware/requirementsJSON

Add in requirements.json creation feature

Author: smlindauer

.github/workflows/build-test-deploy.yml

@@ -12,7 +12,7 @@ jobs:
       - name: Black
         uses: psf/black@stable
         with:
-          src: "./src"
+          src: "./src ./tests"
 
   test:
     name: "Test"

examples/data/hmeqModels/DecisionTreeClassifier/requirements.json

@@ -0,0 +1,14 @@
+[
+    {
+        "step": "install pandas",
+        "command": "pip install pandas==1.2.4"
+    },
+    {
+        "step": "install numpy",
+        "command": "pip install numpy==1.20.1"
+    },
+    {
+        "step": "install sklearn",
+        "command": "pip install sklearn"
+    }
+]

examples/data/hmeqModels/GradientBoosting/requirements.json

@@ -0,0 +1,14 @@
+[
+    {
+        "step": "install pandas",
+        "command": "pip install pandas==1.2.4"
+    },
+    {
+        "step": "install numpy",
+        "command": "pip install numpy==1.20.1"
+    },
+    {
+        "step": "install sklearn",
+        "command": "pip install sklearn"
+    }
+]

examples/data/hmeqModels/RandomForest/requirements.json

@@ -0,0 +1,14 @@
+[
+    {
+        "step": "install pandas",
+        "command": "pip install pandas==1.2.4"
+    },
+    {
+        "step": "install numpy",
+        "command": "pip install numpy==1.20.1"
+    },
+    {
+        "step": "install sklearn",
+        "command": "pip install sklearn"
+    }
+]

examples/pzmmModelImportExample.ipynb

@@ -635,7 +635,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 13,
    "metadata": {
     "Collapsed": "false"
    },
@@ -685,6 +685,9 @@
     "    # Write model metadata to a json file\n",
     "    J.writeFileMetadataJSON(modelPrefix, jPath=zipFolder)\n",
     "\n",
+    "    # Write out requirements.json\n",
+    "    J.createRequirementsJSON(zipFolder)\n",
+    "\n",
     "for (prefix, path) in zip(modelPrefix, zipFolder):\n",
     "    writeJSONFiles(hmeqData, predictorColumns, targetColumn, path, prefix)"
    ]

examples/pzmmModelImportExampleReqJSON.ipynb

@@ -121,7 +121,7 @@ def pzmmImportModel(
         inputDF,
         targetDF,
         predictmethod,
-        metrics=["EM_EVENTPROBABILITY", "EM_CLASSIFICATION"],
+        metrics=None,
         projectVersion="latest",
         modelFileName=None,
         pyPath=None,
@@ -171,7 +171,7 @@ def pzmmImportModel(
             the format() command.
             For example: '{}.predict_proba({})'.
         metrics : string list, optional
-            The scoring metrics for the model. The default is a set of two
+            The scoring metrics for the model. The default is a list of two
             metrics: EM_EVENTPROBABILITY and EM_CLASSIFICATION.
         projectVersion : string, optional
             The project version to import the model in to on SAS Model Manager. The default value
@@ -201,6 +201,10 @@ def pzmmImportModel(
             Model details from an MLFlow model. This dictionary is created by the readMLModelFile function.
             By default None.
         """
+        # Set metrics internal to function call if no value is given
+        if metrics is None:
+            metrics = ["EM_EVENTPROBABILITY", "EM_CLASSIFICATION"]
+
         # Initialize no score code or binary H2O model flags
         noScoreCode = False
         binaryModel = False

src/sasctl/pzmm/importModel.py

@@ -17,7 +17,7 @@ def writeScoreCode(
         modelPrefix,
         predictMethod,
         modelFileName,
-        metrics=["EM_EVENTPROBABILITY", "EM_CLASSIFICATION"],
+        metrics=None,
         pyPath=Path.cwd(),
         threshPrediction=None,
         otherVariable=False,
@@ -120,6 +120,10 @@ def writeScoreCode(
         pickleType : string, optional
             Indicator for MLFlow models, which may pickle by non-standard methods. By default 'pickle'.
         """
+        # Set metrics internal to function call if no value is given
+        if metrics is None:
+            metrics = ["EM_EVENTPROBABILITY", "EM_CLASSIFICATION"]
+
         # Check if binary string model
         if binaryString is not None:
             isBinaryString = True
@@ -147,7 +151,7 @@ def upload_and_copy_score_resources(model, files):
 
         # For SAS Viya 3.5, either return an error or return the model UUID as a string
         if isViya35:
-            if model == None:
+            if model is None:
                 raise ValueError(
                     "The model UUID is required for score code written for"
                     + " SAS Model Manager on SAS Viya 3.5."
@@ -596,6 +600,7 @@ def score{modelPrefix}({inputVarList}):
                 model["scoreCodeType"] = "ds2MultiType"
                 modelRepo.update_model(model)
 
+    @classmethod
     def splitStringColumn(cls, inputSeries, otherVariable):
         """
         Splits a column of string values into a number of new variables equal