no function changes

Author: levkk

pgml-admin/app/serializers.py

@@ -2,8 +2,31 @@
 
 from app.models import Project, Snapshot, Model, Deployment
 
+class SnapshotSerializer(serializers.ModelSerializer):
+    y_column_name = serializers.ListSerializer(child=serializers.CharField())
+
+    class Meta:
+        model = Snapshot
+        fields = [
+            "id",
+            "y_column_name",
+            "test_size",
+            "test_sampling",
+            "status",
+            "columns",
+            "analysis",
+            "sample",
+            "samples",
+            "table_size",
+            "feature_size",
+            "created_at",
+            "updated_at",
+        ]
+
 
 class ModelSerializer(serializers.ModelSerializer):
+    snapshot = SnapshotSerializer()
+
     class Meta:
         model = Model
         fields = [
@@ -27,34 +50,15 @@ class Meta:
         fields = [
             "id",
             "name",
+            "key_metric_name",
+            "key_metric_display_name",
             "objective",
             "created_at",
             "updated_at",
             "models",
         ]
 
 
-class SnapshotSerializer(serializers.ModelSerializer):
-    y_column_name = serializers.ListSerializer(child=serializers.CharField())
-
-    class Meta:
-        model = Snapshot
-        fields = [
-            "id",
-            "y_column_name",
-            "test_size",
-            "test_sampling",
-            "status",
-            "columns",
-            "analysis",
-            "sample",
-            "samples",
-            "table_size",
-            "feature_size",
-            "created_at",
-            "updated_at",
-        ]
-
 
 class DeploymentSerializer(serializers.ModelSerializer):
     class Meta:
@@ -65,8 +69,9 @@ class Meta:
 
 class NewProjectSerializer(serializers.Serializer):
     project_name = serializers.CharField()
-    objective = serializers.CharField()
-    snapshot_id = serializers.IntegerField()
+    objective = serializers.CharField(required=False)
+    relation_name = serializers.CharField(required=False)
+    y_column_name = serializers.ListSerializer(child=serializers.CharField(), required=False)
     algorithms = serializers.ListSerializer(child=serializers.CharField())
 
 

pgml-admin/app/static/js/controllers/new-project.js

@@ -164,16 +164,19 @@ export default class extends Controller {
     }
 
     renderAnalysisResult() {
-      fetch(`/html/snapshots/analysis/?snapshot_id=${this.snapshotData.id}`)
+      const snapshotData = this.projectData.models[0].snapshot
+
+      console.log("Fetching analysis")
+      fetch(`/html/snapshots/analysis/?snapshot_id=${snapshotData.id}`)
       .then(res => res.text())
       .then(html => this.analysisResultTarget.innerHTML = html)
       .then(() => {
         // Render charts
-        for (name in this.snapshotData.columns) {
+        for (name in snapshotData.columns) {
           const sample = JSON.parse(document.getElementById(name).textContent)
-          renderDistribution(name, sample, this.snapshotData.analysis[`${name}_dip`])
+          renderDistribution(name, sample, snapshotData.analysis[`${name}_dip`])
 
-          for (target of this.snapshotData.y_column_name) {
+          for (target of snapshotData.y_column_name) {
             if (target === name)
               continue
 
@@ -182,9 +185,9 @@ export default class extends Controller {
           }
         }
 
-        for (target of this.snapshotData.y_column_name) {
+        for (target of snapshotData.y_column_name) {
           const targetSample = JSON.parse(document.getElementById(target).textContent)
-          renderOutliers(target, targetSample, this.snapshotData.analysis[`${target}_stddev`])
+          renderOutliers(target, targetSample, snapshotData.analysis[`${target}_stddev`])
         }
 
         this.progressBarProgress = 100
@@ -215,52 +218,34 @@ export default class extends Controller {
     createSnapshot(event) {
       event.preventDefault()
 
-      const request = {
-        "relation_name": this.tableName,
-        "y_column_name": Array.from(this.targetNames),
-      }
+      // Train a linear algorithm by default
+      this.algorithmNames.add("linear")
 
       this.nextStep()
 
       // Start the progress bar :)
       this.progressBarProgress = 2
-      this.progressBarInterval = setInterval(this.renderProgressBar.bind(this), 750)
-      console.log("interval set to ", this.progressBarInterval)
+      this.progressBarInterval = setInterval(this.renderProgressBar.bind(this), 850)
 
-      fetch(`/api/snapshots/snapshot/`, {
-        method: "POST",
-        cache: "no-cache",
-        headers: {
-          "Content-Type": "application/json",
-        },
-        redirect: "follow",
-        body: JSON.stringify(request),
-      })
-      .then(res => {
-        if (res.ok) {
-          return res.json()
-        } else {
-          alert("Failed to create snapshot")
-          throw Error("Failed to create snapshot")
-        }
-      })
-      .then(json => {
-        this.snapshotData = json
+      this.createProject(event, false, () => {
         this.renderAnalysisResult()
+        this.algorithmNames.delete("linear")
       })
     }
 
-    createProject(event) {
+    createProject(event, redirect = true, callback = null) {
       event.preventDefault()
 
       const request = {
         "project_name": this.projectName,
         "objective": this.objectiveName,
         "algorithms": Array.from(this.algorithmNames),
-        "snapshot_id": this.snapshotData.id,
+        "relation_name": this.tableName,
+        "y_column_name": Array.from(this.targetNames),
       }
 
-      this.createLoader()
+      if (redirect)
+        this.createLoader()
 
       fetch(`/api/projects/train/`, {
         method: "POST",
@@ -280,7 +265,13 @@ export default class extends Controller {
         }
       })
       .then(json => {
-        window.location.assign(`/projects/${json.id}`);
+        this.projectData = json
+
+        if (redirect)
+          window.location.assign(`/projects/${json.id}`);
+
+        if (callback)
+          callback()
       })
     }
 

pgml-admin/app/templates/projects/new.html

@@ -19,6 +19,23 @@ <h2>Project name</h2>
     </div>
   </section>
 
+  <section data-new-project-target="step" class="hidden" data-step-name="algorithm-type">
+    <h2>Objective</h2>
+    <ol class="objective_list">
+      <li>
+        <a href="#" data-action="click->new-project#selectObjective" data-new-project-target="objective" data-objective="regression">Regression</a>
+      </li>
+      <li>
+        <a href="#" data-action="click->new-project#selectObjective" data-new-project-target="objective" data-objective="classification">Classification</a>
+      </li>
+    </ol>
+
+    <div class="button-container">
+      <button class="next" data-action="click->new-project#nextStep" data-new-project-target="objectiveNameNext" disabled>Next</button>
+      <button class="next" data-action="click->new-project#previousStep">Back</button>
+    </div>
+  </section>
+
   <section data-new-project-target="step" class="hidden" data-step-name="data-source">
     <h2>Data source</h2>
 
@@ -46,14 +63,14 @@ <h2>Target</h2>
     <ol data-new-project-target="trainingLabel" class="object_list snapshot_list"></ol>
 
     <div class="button-container">
-      <button class="next" data-action="click->new-project#createSnapshot" data-new-project-target="analysisNext" disabled>Snapshot data</button>
+      <button class="next" data-action="click->new-project#createSnapshot" data-new-project-target="analysisNext" disabled>Run analysis</button>
       <button class="next" data-action="click->new-project#previousStep">Back</button>
     </div>
   </section>
 
   <section data-new-project-target="step" class="hidden" data-step-name="target">
     <div style="display: flex; justify-content: center; align-items: center;">
-      <p>Creating snapshot, this may take a moment...</p>
+      <p>Performing data analysis, this may take a moment...</p>
     </div>
 
     <div class="progress-bar">
@@ -73,28 +90,10 @@ <h2>Analysis</h2>
     </div>
   </section>
 
-  <section data-new-project-target="step" class="hidden" data-step-name="algorithm-type">
-    <h2>Objective</h2>
-    <ol class="objective_list">
-      <li>
-        <a href="#" data-action="click->new-project#selectObjective" data-new-project-target="objective" data-objective="regression">Regression</a>
-      </li>
-      <li>
-        <a href="#" data-action="click->new-project#selectObjective" data-new-project-target="objective" data-objective="classification">Classification</a>
-      </li>
-    </ol>
-
-    <div class="button-container">
-      <button class="next" data-action="click->new-project#nextStep" data-new-project-target="objectiveNameNext" disabled>Next</button>
-      <button class="next" data-action="click->new-project#previousStep">Back</button>
-    </div>
-  </section>
-
   <section data-new-project-target="step" class="hidden" data-step-name="algorithm">
     <h2>Algorithms</h2>
     <div style="margin-top: 25px" data-new-project-target="algorithmListRegression">
       <ol class="algorithm_list">
-        <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="linear"><span>LinearRegression</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="ridge"><span>Ridge</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="lasso"><span>Lasso</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="elastic_net"><span>ElasticNet</span></a></li>
@@ -127,7 +126,6 @@ <h2>Algorithms</h2>
 
     <div style="margin-top: 25px" data-new-project-target="algorithmListClassification">
       <ol class="algorithm_list">
-        <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="linear"><span>LogisticRegression</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="ridge"><span>RidgeClassifier</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="stochastic_gradient_descent"><span>SGDClassifier</span></a></li>
         <li><a href="#" data-action="click->new-project#selectAlgorithm" data-algorithm="perceptron"><span>Perceptron</span></a></li>

pgml-admin/app/templates/snapshots/analysis.html

@@ -1,3 +1,16 @@
+<div style="margin: 30px 0">
+  <h3><span class="material-symbols-outlined">model_training</span>Linear model</h3>
+  <dl class="model_metrics">
+    {% for key, value in model.metrics.items %}
+    {% if key == "search_results" %}
+    {% else %}
+    <dt>{{ key }}</dt>
+    <dd>{{ value|floatformat:"5"}}</dd>
+    {% endif %}
+    {% endfor %}
+  </dl>
+</div>
+
 <div class="feature_figures">
   <h3><span class="material-symbols-outlined">label_important</span>Labels</h3>
   {% for label in labels %}

pgml-admin/app/views/projects.py

@@ -75,25 +75,34 @@ def train(self, request):
         """Train a new project."""
         serializer = NewProjectSerializer(data=request.data)
         if serializer.is_valid():
+            exists = len(Project.objects.filter(name=serializer.validated_data["project_name"])) > 0
+
             with connection.cursor() as cursor:
-                cursor.execute(
-                    """
-                    SELECT * FROM pgml.train_joint(
-                        project_name => %s,
-                        objective => %s,
-                        snapshot_id => %s,
-                        algorithm => %s
+                if not exists:
+                    cursor.execute(
+                        """
+                        SELECT * FROM pgml.train_joint(
+                            project_name => %s,
+                            objective => %s,
+                            relation_name => %s,
+                            y_column_name => %s,
+                            algorithm => %s
+                        )
+                    """,
+                        [
+                            serializer.validated_data["project_name"],
+                            serializer.validated_data["objective"],
+                            serializer.validated_data["relation_name"],
+                            serializer.validated_data["y_column_name"],
+                            serializer.validated_data["algorithms"][0],
+                        ],
                     )
-                """,
-                    [
-                        serializer.validated_data["project_name"],
-                        serializer.validated_data["objective"],
-                        serializer.validated_data["snapshot_id"],
-                        serializer.validated_data["algorithms"][0],
-                    ],
-                )
-
-                for algorithm in serializer.validated_data["algorithms"][1:]:
+                if exists:
+                    algorithms = serializer.validated_data["algorithms"]
+                else:
+                    algorithms = serializer.validated_data["algorithms"][1:]
+
+                for algorithm in algorithms:
                     cursor.execute(
                         """
                     SELECT * FROM pgml.train(

pgml-admin/app/views/snapshots.py

@@ -11,7 +11,7 @@
 from rest_framework import status
 
 import json
-from app.models import Snapshot, Project
+from app.models import Snapshot, Project, Model
 from app.serializers import SnapshotSerializer, NewSnapshotSerializer
 
 from collections import namedtuple
@@ -108,7 +108,8 @@ def list(self, request):
                     "type": snapshot.columns[column],
                     "samples": list(map(lambda x: x[column], snapshot.sample())),
                 } for column in snapshot.columns.keys() - snapshot.y_column_name
-            ]
+            ],
+            "model": Model.objects.filter(snapshot=snapshot, algorithm_name="linear").first(),
         }
 
         return render(request, "snapshots/analysis.html", context)

pgml-extension/pgml_extension/model.py

@@ -770,7 +770,6 @@ def train(
     objective: str = None,
     relation_name: str = None,
     y_column_name: str = None,
-    snapshot_id: int = None,
     algorithm_name: str = "linear",
     hyperparams: dict = {},
     search: str = None,
@@ -805,18 +804,7 @@ def train(
         raise PgMLException(f"Unknown objective `{objective}`, available options are: regression, classification.")
 
     # Create or use an existing snapshot.
-    #
-    # If a snapshot_id is given, use that specific snapshot.
-    # If a relation name is given, snapshot it.
-    # If none of the above, use the last snapshot created for the project, if any.
-    #
-    if snapshot_id is not None:
-        snapshot = Snapshot.find(snapshot_id)
-        if snapshot is None:
-            raise PgMLException(
-                f"Snapshot with ID {snapshot_id} does not exist."
-            )
-    elif relation_name is None:
+    if relation_name is None:
         snapshot = project.last_snapshot
         if snapshot is None:
             raise PgMLException(