autoforge-engine 0.1.3__tar.gz → 0.1.4__tar.gz

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: autoforge-engine
-Version: 0.1.3
+Version: 0.1.4
 Summary: A transparent, local-first AutoML framework for automated ML pipeline discovery.
 Author: Aditya Kumar Singh
 Author-email: Aditya Kumar Singh <adityasingh45245@gmail.com>
@@ -248,6 +248,9 @@ Pass options such as `task_type="classification"`, `cv=5`, or
 
 Pass a pandas DataFrame to `AutoML.fit`, name the target column, then save the
 fitted pipeline. Replace the example path with the path to your own dataset.
+`fit` prints a report with dataset details, all evaluated model metrics, the
+selected pipeline, and experiment identifiers by default. Pass
+`print_report=False` to suppress it.
 
 ```python
 from pathlib import Path

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/README.md

@@ -216,6 +216,9 @@ Pass options such as `task_type="classification"`, `cv=5`, or
 
 Pass a pandas DataFrame to `AutoML.fit`, name the target column, then save the
 fitted pipeline. Replace the example path with the path to your own dataset.
+`fit` prints a report with dataset details, all evaluated model metrics, the
+selected pipeline, and experiment identifiers by default. Pass
+`print_report=False` to suppress it.
 
 ```python
 from pathlib import Path

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/autoforge_engine.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: autoforge-engine
-Version: 0.1.3
+Version: 0.1.4
 Summary: A transparent, local-first AutoML framework for automated ML pipeline discovery.
 Author: Aditya Kumar Singh
 Author-email: Aditya Kumar Singh <adityasingh45245@gmail.com>
@@ -248,6 +248,9 @@ Pass options such as `task_type="classification"`, `cv=5`, or
 
 Pass a pandas DataFrame to `AutoML.fit`, name the target column, then save the
 fitted pipeline. Replace the example path with the path to your own dataset.
+`fit` prints a report with dataset details, all evaluated model metrics, the
+selected pipeline, and experiment identifiers by default. Pass
+`print_report=False` to suppress it.
 
 ```python
 from pathlib import Path

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/modelforge/automl.py

@@ -233,6 +233,7 @@ class AutoML:
         task_type: str | None = None,
         model_names: list[str] | None = None,
         excluded_columns: list[str] | None = None,
+        print_report: bool = True,
     ) -> dict[str, Any]:
         """
         Run the complete ModelForge AutoML workflow.
@@ -241,8 +242,12 @@ class AutoML:
         recorded after successful or failed execution.
 
         Reproducibility metadata is captured for every run.
+        Set print_report=False to suppress the completion report.
         """
 
+        if not isinstance(print_report, bool):
+            raise TypeError("print_report must be a boolean.")
+
         target = (
             target
             if target is not None
@@ -357,6 +362,9 @@ class AutoML:
 
             self.result = result
 
+            if print_report:
+                _print_automl_report(self, result)
+
             return result
 
         except Exception as exc:
@@ -1487,14 +1495,13 @@ def automl(
     """Fit AutoML, print a run report, and return the fitted engine."""
 
     engine = AutoML(**options)
-    result = engine.fit(
+    engine.fit(
         data=data,
         target=target,
         task_type=task_type,
         model_names=model_names,
         excluded_columns=excluded_columns,
     )
-    _print_automl_report(engine, result)
     return engine
 
 
@@ -1508,6 +1515,7 @@ def _print_automl_report(
     profile = result["profile"]
     audit = result["audit"]
     ranking = result["ranking"]
+    target_info = result["target"]
     missing_values = sum(
         column.get("missing_values", 0)
         for column in profile["column_info"].values()
@@ -1525,9 +1533,8 @@ def _print_automl_report(
     dataset_table.add_column("Value", style="white")
     dataset_table.add_row("Rows", f"{profile['rows']:,}")
     dataset_table.add_row("Columns", f"{profile['columns']:,}")
-    dataset_table.add_row("Features", f"{profile['columns'] - 1:,}")
-    dataset_table.add_row("Target", str(result["target"]["target"]))
-    dataset_table.add_row("Task", str(result["target"]["task_type"]))
+    dataset_table.add_row("Target", str(target_info["target"]))
+    dataset_table.add_row("Task", str(target_info["task_type"]))
     dataset_table.add_row("Missing values", f"{missing_values:,}")
     dataset_table.add_row(
         "Duplicate rows",
@@ -1538,43 +1545,56 @@ def _print_automl_report(
         Panel(dataset_table, title="Dataset", border_style="cyan")
     )
 
-    metric_column = (
-        "cv_mean_r2"
-        if engine.task_type == "regression"
-        else "cv_mean_f1"
-    )
-    metric_label = "CV R2" if engine.task_type == "regression" else "CV F1"
-    ranking_columns = [
-        ("rank", "#"),
-        ("model", "Model"),
-        ("overall_score", "Overall"),
-        (metric_column, metric_label),
-        ("speed_score", "Speed"),
-        ("status", "Status"),
+    metric_columns = [
+        column
+        for column in ranking.columns
+        if column.startswith("cv_mean_")
+        and not column.endswith("_seconds")
     ]
     ranking_table = Table(box=None, padding=(0, 1))
-    for column, label in ranking_columns:
-        if column in ranking.columns:
-            ranking_table.add_column(label)
-
-    for _, row in ranking.head(10).iterrows():
-        values = []
-        for column, _ in ranking_columns:
-            if column not in ranking.columns:
-                continue
+    ranking_table.add_column("#", justify="right", no_wrap=True)
+    ranking_table.add_column("Model", overflow="fold")
+    ranking_table.add_column("CV Metrics", overflow="fold")
+
+    for _, row in ranking.sort_values(
+        "rank",
+        na_position="last",
+    ).iterrows():
+        metric_details = []
+        for column in metric_columns:
+            metric_label = column.removeprefix("cv_mean_")
+            metric_label = metric_label.replace("_", " ").upper()
+            metric_label = metric_label.replace("ROC AUC", "ROC-AUC")
             value = row[column]
-            if isinstance(value, (int, float)):
-                values.append(f"{value:.4f}")
-            else:
-                values.append(str(value))
-        ranking_table.add_row(*values)
+            formatted_value = (
+                "N/A" if pd.isna(value) else f"{value:.4f}"
+            )
+            metric_details.append(
+                f"CV {metric_label}: {formatted_value}"
+            )
+        if "overall_score" in ranking.columns:
+            metric_details.append(
+                f"Overall: {row['overall_score']:.4f}"
+            )
+        if "status" in ranking.columns:
+            metric_details.append(f"Status: {row['status']}")
+        model_name = engine.registry.get(str(row["model"])).name
+        rank = row.get("rank")
+        rank_label = "N/A" if pd.isna(rank) else str(int(rank))
+        ranking_table.add_row(
+            rank_label,
+            model_name,
+            " | ".join(metric_details),
+        )
     console.print(
         Panel(ranking_table, title="Model Ranking", border_style="green")
     )
 
-    best_row = ranking.loc[
-        ranking["model"] == engine.best_model
-    ].iloc[0]
+    best_model = result.get("best_model", engine.best_model)
+    best_matches = ranking.loc[
+        ranking["model"].astype(str) == str(best_model)
+    ]
+    best_row = best_matches.iloc[0] if not best_matches.empty else None
     best_table = Table(
         show_header=False,
         box=None,
@@ -1582,13 +1602,25 @@ def _print_automl_report(
     )
     best_table.add_column("Property", style="cyan")
     best_table.add_column("Value", style="white")
-    best_table.add_row("Model", str(engine.best_model))
-    best_table.add_row(
-        "Overall score",
-        f"{best_row['overall_score']:.4f}",
-    )
-    if metric_column in ranking.columns:
-        best_table.add_row(metric_label, f"{best_row[metric_column]:.4f}")
+    best_model_name = engine.registry.get(str(best_model)).name
+    best_table.add_row("Model", best_model_name)
+    if best_row is not None:
+        if "rank" in ranking.columns:
+            best_table.add_row("Rank", str(int(best_row["rank"])))
+        if "overall_score" in ranking.columns:
+            best_table.add_row(
+                "Overall score",
+                f"{best_row['overall_score']:.4f}",
+            )
+        for column in metric_columns:
+            metric_label = column.removeprefix("cv_mean_")
+            metric_label = metric_label.replace("_", " ").upper()
+            metric_label = metric_label.replace("ROC AUC", "ROC-AUC")
+            value = best_row[column]
+            best_table.add_row(
+                f"CV {metric_label}",
+                "N/A" if pd.isna(value) else f"{value:.4f}",
+            )
     best_table.add_row("Objective", str(engine.objective))
     best_table.add_row("Cross-validation", f"{engine.cv} folds")
     best_table.add_row("Holdout", f"{engine.test_size:.0%}")
@@ -1608,6 +1640,51 @@ def _print_automl_report(
         Panel(best_table, title="Best Model & Settings", border_style="yellow")
     )
 
+    pipeline_table = Table(
+        show_header=False,
+        box=None,
+        padding=(0, 1),
+    )
+    pipeline_table.add_column("Property", style="cyan")
+    pipeline_table.add_column("Value", style="white")
+    pipeline = result.get("best_pipeline")
+    pipeline_steps = getattr(pipeline, "steps", [])
+    for step_name, step in pipeline_steps:
+        pipeline_table.add_row(
+            str(step_name),
+            type(step).__name__,
+        )
+        for transformer_name, transformer, _ in getattr(
+            step,
+            "transformers_",
+            [],
+        ):
+            if isinstance(transformer, str):
+                description = transformer
+            else:
+                nested_steps = getattr(transformer, "steps", None)
+                if nested_steps:
+                    description = " -> ".join(
+                        type(component).__name__
+                        for _, component in nested_steps
+                    )
+                else:
+                    description = type(transformer).__name__
+            pipeline_table.add_row(
+                f"  {transformer_name}",
+                description,
+            )
+    for name, value in result.get("feature_selection", {}).items():
+        pipeline_table.add_row(
+            name.replace("_", " ").title(),
+            "disabled" if value is None else str(value),
+        )
+    pipeline_table.add_row("Cross-validation", f"{engine.cv}-fold")
+    pipeline_table.add_row("Random state", str(engine.random_state))
+    console.print(
+        Panel(pipeline_table, title="Pipeline Information", border_style="cyan")
+    )
+
     audit_table = Table(box=None, padding=(0, 1))
     audit_table.add_column("Severity", style="yellow")
     audit_table.add_column("Finding")

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/modelforge/cli.py

@@ -377,6 +377,7 @@ def train(
                 task_type=resolved_task_type,
                 model_names=selected_models,
                 excluded_columns=excluded_columns,
+                print_report=False,
             )
 
         model_path = automl.save(

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/modelforge/model_registry.py

@@ -487,10 +487,9 @@ class ModelRegistry:
                 task_type="classification",
                 category="svm",
                 requires_scaling=True,
-                supports_probability=True,
+                supports_probability=False,
                 default_params={
                     "kernel": "rbf",
-                    "probability": True,
                 },
                 hyperparameter_space={
                     "C": [

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "autoforge-engine"
-version = "0.1.3"
+version = "0.1.4"
 description = "A transparent, local-first AutoML framework for automated ML pipeline discovery."
 readme = "README.md"
 requires-python = ">=3.11"
@@ -54,4 +54,5 @@ modelforge = "modelforge.cli:app"
 include = ["modelforge*"]
 
 [tool.pytest.ini_options]
-testpaths = ["tests"]
+testpaths = ["tests"]
+pythonpath = ["."]

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/tests/test_automl.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import json
+import warnings
 
 import numpy as np
 import pandas as pd
@@ -100,7 +101,7 @@ def test_automl_convenience_function_prints_report(
         data_path,
         "target",
         task_type="regression",
-        model_names=["linear_regression"],
+        model_names=["linear_regression", "ridge"],
         cv=2,
         experiment_directory=tmp_path / "experiments",
     )
@@ -113,7 +114,86 @@ def test_automl_convenience_function_prints_report(
     assert "Best Model & Settings" in output
     assert "Data Quality" in output
     assert "Run Information" in output
+    assert "Pipeline Information" in output
     assert "AUTOFORGE COMPLETE" in output
+    assert f"{fitted.result['profile']['rows']:,}" in output
+    assert f"{fitted.result['profile']['columns']:,}" in output
+    assert fitted.result["target"]["target"] in output
+    assert fitted.result["target"]["task_type"] in output
+    assert fitted.experiment_id in output
+    for model in fitted.result["ranking"]["model"]:
+        assert fitted.registry.get(model).name in output
+    metric_columns = [
+        column
+        for column in fitted.result["ranking"].columns
+        if column.startswith("cv_mean_")
+        and not column.endswith("_seconds")
+    ]
+    assert metric_columns
+    for column in metric_columns:
+        metric_label = column.removeprefix("cv_mean_")
+        metric_label = metric_label.replace("_", " ").upper()
+        metric_label = metric_label.replace("ROC AUC", "ROC-AUC")
+        assert f"CV {metric_label}" in output
+
+
+def test_automl_fit_prints_report_by_default(
+    capsys,
+    tmp_path,
+):
+    data = regression_data()
+    engine = AutoML(
+        cv=2,
+        experiment_directory=tmp_path / "experiments",
+    )
+
+    result = engine.fit(
+        data,
+        target="target",
+        task_type="regression",
+        model_names=["linear_regression", "ridge"],
+    )
+
+    output = capsys.readouterr().out
+    assert f"{result['profile']['rows']:,}" in output
+    assert f"{result['profile']['columns']:,}" in output
+    assert result["target"]["target"] in output
+    assert result["target"]["task_type"] in output
+    assert result["experiment_id"] in output
+    for model in result["ranking"]["model"]:
+        assert engine.registry.get(model).name in output
+    assert "Pipeline Information" in output
+
+
+def test_classification_report_shows_metrics_for_all_models(
+    capsys,
+    tmp_path,
+):
+    engine = AutoML(
+        cv=2,
+        experiment_directory=tmp_path / "experiments",
+    )
+
+    with warnings.catch_warnings():
+        warnings.simplefilter("error", FutureWarning)
+        result = engine.fit(
+            classification_data(),
+            target="target",
+            task_type="classification",
+            model_names=["logistic_regression", "svc"],
+        )
+
+    output = capsys.readouterr().out
+    assert "CV F1" in output
+    assert "CV ROC-AUC" in output
+    assert engine.registry.get(result["best_model"]).name in output
+    assert result["experiment_id"] in output
+    for model in result["ranking"]["model"]:
+        assert engine.registry.get(model).name in output
+    for column in ("cv_mean_f1", "cv_mean_roc_auc"):
+        for value in result["ranking"][column]:
+            rendered_value = "N/A" if pd.isna(value) else f"{value:.4f}"
+            assert rendered_value in output
 
 
 def test_automl_rejects_invalid_variance_threshold():

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/tests/test_cli.py

@@ -1,3 +1,5 @@
+import re
+
 from typer.testing import CliRunner
 
 from modelforge.cli import app
@@ -244,7 +246,17 @@ def test_train_config_help():
     )
 
     assert result.exit_code == 0
-    assert "--config" in result.stdout
+
+    # Typer/Rich can add ANSI escape sequences to
+    # help output in CI environments. Remove them
+    # before checking the actual CLI text.
+    clean_output = re.sub(
+        r"\x1b\[[0-9;]*m",
+        "",
+        result.stdout,
+    )
+
+    assert "--config" in clean_output
 
 
 def test_train_cli_target_overrides_config(

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/tests/test_config_automl_integration.py

@@ -124,7 +124,8 @@ def test_fit_can_use_target_from_config(monkeypatch):
                 "price": [1, 2, 3],
                 "feature": [4, 5, 6],
             }
-        )
+        ),
+        print_report=False,
     )
 
     assert captured["target"] == "price"
@@ -187,6 +188,7 @@ def test_explicit_fit_values_override_config():
             task_type="regression",
             model_names=["ridge"],
             excluded_columns=["explicit_column"],
+            print_report=False,
         )
     finally:
         monkeypatch.undo()

{autoforge_engine-0.1.3 → autoforge_engine-0.1.4}/tests/test_model_registry.py

@@ -153,6 +153,20 @@ def test_logistic_regression_requires_scaling():
     assert spec.supports_probability is True
 
 
+def test_svc_uses_decision_scores_without_probability_estimates():
+    registry = ModelRegistry()
+
+    spec = registry.get("svc")
+    model = registry.create("svc")
+
+    assert spec.supports_probability is False
+    assert "probability" not in spec.default_params
+    assert model.get_params(deep=False)["probability"] in {
+        False,
+        "deprecated",
+    }
+
+
 def test_tree_models_do_not_require_scaling():
     registry = ModelRegistry()