wizata-dsapi 2.0.0.dev26__tar.gz → 2.0.0.dev27__tar.gz

{wizata_dsapi-2.0.0.dev26/wizata_dsapi.egg-info → wizata_dsapi-2.0.0.dev27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: wizata_dsapi
-Version: 2.0.0.dev26
+Version: 2.0.0.dev27
 Summary: Wizata Data Science Toolkit
 Author: Wizata S.A.
 Author-email: info@wizata.com

wizata_dsapi-2.0.0.dev27/wizata_dsapi/plots/__init__.py

@@ -0,0 +1,2 @@
+from .common import ts_chart, confusion_matrix, r_squared, anomalies_chart, parallel_coordinates, data_table, setpoint_recommendation, feature_importance
+from . import theme

{wizata_dsapi-2.0.0.dev26 → wizata_dsapi-2.0.0.dev27}/wizata_dsapi/plots/common.py

@@ -220,6 +220,147 @@ def parallel_coordinates(context: wizata_dsapi.Context):
     )
 
 
+def setpoint_recommendation(context: wizata_dsapi.Context):
+    """Compare current vs recommended setpoint values as a Wizata-themed table.
+
+    Auto-pairs columns: for every column ending in '_recommended', looks for a matching column
+    with the same name minus the suffix. Validates that the base column corresponds to a datapoint
+    with BusinessType.SET_POINTS via context.datapoints (defense against accidental pairings).
+    Uses the **last row** of the dataframe for current and recommended values (most recent state).
+
+    Output columns: Setpoint | Current | Recommended | Δ | Δ (%)
+    """
+    df = context.dataframe
+    datapoints = context.datapoints or {}
+
+    if df is None or len(df) == 0:
+        raise ValueError("setpoint_recommendation: dataframe is empty")
+
+    suffix = "_recommended"
+    pairs = []
+    for col in df.columns:
+        if not isinstance(col, str) or not col.endswith(suffix):
+            continue
+        base = col[: -len(suffix)]
+        if base not in df.columns:
+            continue
+        # Safety check: only pair if the base column is a declared setpoint datapoint.
+        # If datapoints are not populated (e.g. local test), fall back to pairing by name.
+        if datapoints:
+            dp = datapoints.get(base)
+            if dp is None or dp.business_type != wizata_dsapi.BusinessType.SET_POINTS:
+                continue
+        pairs.append((base, col))
+
+    if not pairs:
+        raise ValueError(
+            "setpoint_recommendation: no <col>_recommended / setpoint pairs found in the dataframe"
+        )
+
+    last = df.iloc[-1]
+    rows = []
+    for base, rec in pairs:
+        current = last[base]
+        recommended = last[rec]
+        delta = recommended - current
+        pct = (delta / current * 100.0) if current not in (0, 0.0) and not pd.isna(current) else float("nan")
+        rows.append({
+            "Setpoint": base,
+            "Current": current,
+            "Recommended": recommended,
+            "Δ": delta,
+            "Δ (%)": pct,
+        })
+
+    out = pd.DataFrame(rows)
+
+    # Formatting — round numerics for readability
+    for col in ("Current", "Recommended", "Δ", "Δ (%)"):
+        out[col] = out[col].apply(lambda v: "—" if pd.isna(v) else f"{v:.3f}")
+
+    n_rows = len(out)
+    cell_bg = [TABLE_CELL_BG[i % 2] for i in range(n_rows)]
+
+    fig = go.Figure(data=[go.Table(
+        header=dict(
+            values=[f"<b>{c}</b>" for c in out.columns],
+            fill_color=TABLE_HEADER_BG,
+            font=dict(family=FONT_FAMILY, size=FONT_SIZE, color=TABLE_HEADER_TEXT),
+            align="left",
+            line_color=TABLE_LINE_COLOR,
+            height=32,
+        ),
+        cells=dict(
+            values=[out[c].tolist() for c in out.columns],
+            fill_color=[cell_bg],
+            font=dict(family=FONT_FAMILY, size=FONT_SIZE, color=TABLE_CELL_TEXT),
+            align="left",
+            line_color=TABLE_LINE_COLOR,
+            height=28,
+        ),
+    )])
+
+    fig.update_layout(margin=dict(l=0, r=0, t=0, b=0))
+    apply_theme(fig)
+
+    context.set_plot(figure=fig, name="setpoint_recommendation")
+
+
+def feature_importance(context: wizata_dsapi.Context):
+    """Plot the top-N feature importances of a trained model as a horizontal bar chart.
+
+    Scans context.models for a model whose underlying trained_model exposes `feature_importances_`
+    (RandomForestRegressor, RandomForestClassifier, GradientBoostingClassifier, etc.). Uses
+    `ml_model.input_columns` for feature labels. Bars are sorted ascending so the most important
+    feature is at the top when rendered.
+
+    Property: top_n (default 15) — maximum number of features to display.
+    """
+    top_n = int(context.properties.get("top_n", 15)) if "top_n" in context.properties else 15
+
+    selected = None
+    for model in (context.models or {}).values():
+        trained = getattr(model, "trained_model", None)
+        if trained is not None and hasattr(trained, "feature_importances_"):
+            selected = model
+            break
+
+    if selected is None:
+        raise ValueError(
+            "feature_importance: no trained model with feature_importances_ found in context.models "
+            "(supported by RandomForest, GradientBoosting, and similar tree-based models)"
+        )
+
+    importances = np.asarray(selected.trained_model.feature_importances_)
+    feature_names = list(selected.input_columns) if selected.input_columns is not None else \
+        [f"f{i}" for i in range(len(importances))]
+
+    if len(feature_names) != len(importances):
+        raise ValueError(
+            f"feature_importance: feature_importances_ length {len(importances)} does not match "
+            f"input_columns length {len(feature_names)}"
+        )
+
+    fi = pd.DataFrame({"feature": feature_names, "importance": importances})
+    fi = fi.sort_values("importance", ascending=True).tail(top_n)
+
+    fig = go.Figure(go.Bar(
+        x=fi["importance"],
+        y=fi["feature"],
+        orientation="h",
+        marker_color=SERIES_COLORS[0],
+    ))
+    fig.update_layout(
+        title=f"Top-{min(top_n, len(fi))} Feature Importances",
+        xaxis=dict(title="Importance"),
+        yaxis=dict(title=""),
+        margin=dict(l=200, r=20, t=40, b=40),
+    )
+    apply_theme(fig)
+
+    context.set_plot(figure=fig, name="feature_importance")
+
+
 def data_table(context: wizata_dsapi.Context):
     """Render the dataframe as a styled table with Wizata theme colors."""
     df = context.dataframe.copy()

wizata_dsapi-2.0.0.dev27/wizata_dsapi/version.py

@@ -0,0 +1 @@
+__version__ = "2.0.0.dev27"

{wizata_dsapi-2.0.0.dev26 → wizata_dsapi-2.0.0.dev27/wizata_dsapi.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: wizata_dsapi
-Version: 2.0.0.dev26
+Version: 2.0.0.dev27
 Summary: Wizata Data Science Toolkit
 Author: Wizata S.A.
 Author-email: info@wizata.com

wizata_dsapi-2.0.0.dev26/wizata_dsapi/plots/__init__.py

@@ -1,2 +0,0 @@
-from .common import ts_chart, confusion_matrix, r_squared, anomalies_chart, parallel_coordinates, data_table
-from . import theme

wizata_dsapi-2.0.0.dev26/wizata_dsapi/version.py

@@ -1 +0,0 @@
-__version__ = "2.0.0.dev26"