churnkit 0.76.0a3__py3-none-any.whl → 0.76.1a2__py3-none-any.whl

{churnkit-0.76.0a3.data → churnkit-0.76.1a2.data}/data/share/churnkit/exploration_notebooks/01a_temporal_deep_dive.ipynb

@@ -96,22 +96,26 @@
    "outputs": [],
    "source": [
     "from customer_retention.analysis.notebook_progress import track_and_export_previous\n",
+    "\n",
     "track_and_export_previous(\"01a_temporal_deep_dive.ipynb\")\n",
     "\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "import plotly.graph_objects as go\n",
+    "from plotly.subplots import make_subplots\n",
+    "\n",
     "from customer_retention.analysis.auto_explorer import ExplorationFindings\n",
     "from customer_retention.analysis.visualization import ChartBuilder, display_figure, display_table\n",
     "from customer_retention.core.config.column_config import ColumnType, DatasetGranularity\n",
+    "from customer_retention.core.config.experiments import (\n",
+    "    FINDINGS_DIR,\n",
+    ")\n",
     "from customer_retention.stages.profiling import (\n",
-    "    TimeSeriesProfiler, TimeSeriesProfile,\n",
+    "    DistributionAnalyzer,\n",
+    "    TimeSeriesProfiler,\n",
+    "    TransformationType,\n",
     "    TypeDetector,\n",
-    "    DistributionAnalyzer, TransformationType,\n",
-    ")\n",
-    "import pandas as pd\n",
-    "import numpy as np\n",
-    "import plotly.graph_objects as go\n",
-    "import plotly.express as px\n",
-    "from plotly.subplots import make_subplots\n",
-    "from customer_retention.core.config.experiments import FINDINGS_DIR, EXPERIMENTS_DIR, OUTPUT_DIR, setup_experiments_structure\n"
+    ")\n"
    ]
   },
   {
@@ -141,7 +145,6 @@
     "# FINDINGS_PATH = \"../experiments/findings/transactions_abc123_findings.yaml\"\n",
     "\n",
     "# Option 2: Auto-discover findings files\n",
-    "from pathlib import Path\n",
     "\n",
     "# FINDINGS_DIR imported from customer_retention.core.config.experiments\n",
     "\n",
@@ -238,7 +241,7 @@
    },
    "outputs": [],
    "source": [
-    "from customer_retention.stages.temporal import load_data_with_snapshot_preference, TEMPORAL_METADATA_COLS\n",
+    "from customer_retention.stages.temporal import TEMPORAL_METADATA_COLS, load_data_with_snapshot_preference\n",
     "\n",
     "df, data_source = load_data_with_snapshot_preference(findings, output_dir=str(FINDINGS_DIR))\n",
     "charts = ChartBuilder()\n",
@@ -280,7 +283,7 @@
     "    # Manual configuration - uncomment and set if auto-detection failed\n",
     "    # ENTITY_COLUMN = \"customer_id\"\n",
     "    # TIME_COLUMN = \"event_date\"\n",
-    "    \n",
+    "\n",
     "    # Try auto-detection\n",
     "    detector = TypeDetector()\n",
     "    granularity = detector.detect_granularity(df)\n",
@@ -346,17 +349,17 @@
     "print(\"=\"*70)\n",
     "print(\"TIME SERIES PROFILE SUMMARY\")\n",
     "print(\"=\"*70)\n",
-    "print(f\"\\n\\U0001f4ca Dataset Overview:\")\n",
+    "print(\"\\n\\U0001f4ca Dataset Overview:\")\n",
     "print(f\"   Total Events: {ts_profile.total_events:,}\")\n",
     "print(f\"   Unique Entities: {ts_profile.unique_entities:,}\")\n",
     "print(f\"   Avg Events/Entity: {ts_profile.events_per_entity.mean:.1f}\")\n",
     "print(f\"   Time Span: {ts_profile.time_span_days:,} days ({ts_profile.time_span_days/365:.1f} years)\")\n",
     "\n",
-    "print(f\"\\n\\U0001f4c5 Date Range:\")\n",
+    "print(\"\\n\\U0001f4c5 Date Range:\")\n",
     "print(f\"   First Event: {ts_profile.first_event_date}\")\n",
     "print(f\"   Last Event: {ts_profile.last_event_date}\")\n",
     "\n",
-    "print(f\"\\n\\u23f1\\ufe0f  Inter-Event Timing:\")\n",
+    "print(\"\\n\\u23f1\\ufe0f  Inter-Event Timing:\")\n",
     "if ts_profile.avg_inter_event_days is not None:\n",
     "    print(f\"   Avg Days Between Events: {ts_profile.avg_inter_event_days:.1f}\")\n",
     "else:\n",
@@ -791,7 +794,7 @@
    },
    "outputs": [],
    "source": [
-    "print(f\"Coverage Summary:\")\n",
+    "print(\"Coverage Summary:\")\n",
     "print(f\"  Time span: {coverage_result.time_span_days:,} days \"\n",
     "      f\"({coverage_result.first_event.strftime('%Y-%m-%d')} to {coverage_result.last_event.strftime('%Y-%m-%d')})\")\n",
     "print(f\"  Volume trend: {coverage_result.volume_trend} ({coverage_result.volume_change_pct:+.0%})\")\n",
@@ -799,11 +802,11 @@
     "      + (f\" ({sum(g.duration_days for g in coverage_result.gaps):.0f} total days)\" if coverage_result.gaps else \"\"))\n",
     "\n",
     "if coverage_result.recommendations:\n",
-    "    print(f\"\\nRecommendations:\")\n",
+    "    print(\"\\nRecommendations:\")\n",
     "    for rec in coverage_result.recommendations:\n",
     "        print(f\"  -> {rec}\")\n",
     "else:\n",
-    "    print(f\"\\nNo coverage issues detected — data is suitable for all candidate windows.\")"
+    "    print(\"\\nNo coverage issues detected — data is suitable for all candidate windows.\")"
    ]
   },
   {
@@ -843,7 +846,7 @@
     "if drift.recommended_training_start:\n",
     "    print(f\"  Recommended training start: {drift.recommended_training_start.strftime('%Y-%m-%d')}\")\n",
     "\n",
-    "print(f\"\\nRationale:\")\n",
+    "print(\"\\nRationale:\")\n",
     "for r in drift.rationale:\n",
     "    print(f\"  -> {r}\")"
    ]
@@ -904,7 +907,7 @@
     "\n",
     "if inter_event_times:\n",
     "    inter_event_series = pd.Series(inter_event_times)\n",
-    "    \n",
+    "\n",
     "    print(\"\\u23f1\\ufe0f  Inter-Event Time Distribution (days):\")\n",
     "    print(f\"   Min: {inter_event_series.min():.2f}\")\n",
     "    print(f\"   25th percentile: {inter_event_series.quantile(0.25):.2f}\")\n",
@@ -912,14 +915,14 @@
     "    print(f\"   Mean: {inter_event_series.mean():.2f}\")\n",
     "    print(f\"   75th percentile: {inter_event_series.quantile(0.75):.2f}\")\n",
     "    print(f\"   Max: {inter_event_series.max():.2f}\")\n",
-    "    \n",
+    "\n",
     "    # Histogram\n",
     "    fig = go.Figure()\n",
-    "    \n",
+    "\n",
     "    # Cap at 99th percentile for visualization\n",
     "    cap = inter_event_series.quantile(0.99)\n",
     "    display_data = inter_event_series[inter_event_series <= cap]\n",
-    "    \n",
+    "\n",
     "    fig.add_trace(go.Histogram(\n",
     "        x=display_data,\n",
     "        nbinsx=50,\n",
@@ -927,11 +930,11 @@
     "        marker_color=\"coral\",\n",
     "        opacity=0.7\n",
     "    ))\n",
-    "    \n",
+    "\n",
     "    fig.add_vline(x=inter_event_series.median(), line_dash=\"solid\", line_color=\"green\",\n",
     "                  annotation_text=f\"Median: {inter_event_series.median():.1f} days\",\n",
     "                  annotation_position=\"top right\")\n",
-    "    \n",
+    "\n",
     "    fig.update_layout(\n",
     "        title=f\"Inter-Event Time Distribution (capped at {cap:.0f} days = 99th percentile)\",\n",
     "        xaxis_title=\"Days Between Events\",\n",
@@ -978,7 +981,7 @@
     "    if skew_ratio > 1.5:\n",
     "        print(f\"  Distribution is heavily right-skewed (mean/median = {skew_ratio:.2f})\")\n",
     "        print(f\"  -> Most entities engage frequently (median {median_iet:.0f}d between events)\")\n",
-    "        print(f\"  -> A long tail of entities has very infrequent engagement\")\n",
+    "        print(\"  -> A long tail of entities has very infrequent engagement\")\n",
     "    elif skew_ratio > 1.2:\n",
     "        print(f\"  Distribution is moderately right-skewed (mean/median = {skew_ratio:.2f})\")\n",
     "        print(f\"  -> Typical engagement every {median_iet:.0f} days, with some long gaps\")\n",
@@ -988,11 +991,11 @@
     "\n",
     "    print(f\"\\n  Spread: IQR = {iqr:.0f} days (Q25={q25:.0f}d to Q75={q75:.0f}d)\")\n",
     "    if iqr > median_iet:\n",
-    "        print(f\"  -> High variability (IQR > median) — entities have inconsistent timing\")\n",
+    "        print(\"  -> High variability (IQR > median) — entities have inconsistent timing\")\n",
     "    else:\n",
-    "        print(f\"  -> Moderate variability — most entities follow a similar cadence\")\n",
+    "        print(\"  -> Moderate variability — most entities follow a similar cadence\")\n",
     "\n",
-    "    print(f\"\\nRecommendations:\")\n",
+    "    print(\"\\nRecommendations:\")\n",
     "    # Window alignment\n",
     "    window_map = [(1, \"24h\"), (7, \"7d\"), (14, \"14d\"), (30, \"30d\"),\n",
     "                  (90, \"90d\"), (180, \"180d\"), (365, \"365d\")]\n",
@@ -1000,7 +1003,7 @@
     "    if aligned:\n",
     "        aligned_str = \", \".join(w for _, w in aligned)\n",
     "        print(f\"  -> Windows aligned with median inter-event time: {aligned_str}\")\n",
-    "        print(f\"     These capture ~2 events per entity on average\")\n",
+    "        print(\"     These capture ~2 events per entity on average\")\n",
     "    else:\n",
     "        print(f\"  -> Median inter-event ({median_iet:.0f}d) does not align with standard windows\")\n",
     "\n",
@@ -1010,11 +1013,11 @@
     "        print(f\"  -> 30d window captures only ~{events_in_30d:.1f} events/entity — \"\n",
     "              f\"consider longer windows (90d+) for meaningful aggregations\")\n",
     "    if median_iet < 7:\n",
-    "        print(f\"  -> High frequency engagement — 7d and 24h windows will be rich with signal\")\n",
+    "        print(\"  -> High frequency engagement — 7d and 24h windows will be rich with signal\")\n",
     "\n",
     "    if skew_ratio > 1.5:\n",
-    "        print(f\"  -> Consider log-transforming inter-event time as a feature \"\n",
-    "              f\"(reduces right-skew impact on models)\")\n"
+    "        print(\"  -> Consider log-transforming inter-event time as a feature \"\n",
+    "              \"(reduces right-skew impact on models)\")\n"
    ]
   },
   {
@@ -1061,13 +1064,13 @@
     "# Use framework's DistributionAnalyzer for comprehensive analysis\n",
     "analyzer = DistributionAnalyzer()\n",
     "\n",
-    "numeric_cols = [n for n, c in findings.columns.items() \n",
+    "numeric_cols = [n for n, c in findings.columns.items()\n",
     "                if c.inferred_type.value in ('numeric_continuous', 'numeric_discrete')\n",
     "                and n not in [ENTITY_COLUMN, TIME_COLUMN] and n not in TEMPORAL_METADATA_COLS]\n",
     "\n",
     "# Analyze all numeric columns using the framework\n",
     "analyses = analyzer.analyze_dataframe(df, numeric_cols)\n",
-    "recommendations = {col: analyzer.recommend_transformation(analysis) \n",
+    "recommendations = {col: analyzer.recommend_transformation(analysis)\n",
     "                   for col, analysis in analyses.items()}\n",
     "\n",
     "# Human-readable transformation names\n",
@@ -1092,25 +1095,25 @@
     "    col_info = findings.columns[col_name]\n",
     "    analysis = analyses.get(col_name)\n",
     "    rec = recommendations.get(col_name)\n",
-    "    \n",
+    "\n",
     "    print(f\"\\n{'='*70}\")\n",
     "    print(f\"Column: {col_name}\")\n",
     "    print(f\"Type: {col_info.inferred_type.value} (Confidence: {col_info.confidence:.0%})\")\n",
-    "    print(f\"-\" * 70)\n",
-    "    \n",
+    "    print(\"-\" * 70)\n",
+    "\n",
     "    if analysis:\n",
-    "        print(f\"📊 Distribution Statistics:\")\n",
+    "        print(\"📊 Distribution Statistics:\")\n",
     "        print(f\"   Mean: {analysis.mean:.3f}  |  Median: {analysis.median:.3f}  |  Std: {analysis.std:.3f}\")\n",
     "        print(f\"   Range: [{analysis.min_value:.3f}, {analysis.max_value:.3f}]\")\n",
     "        print(f\"   Percentiles: 1%={analysis.percentiles['p1']:.3f}, 25%={analysis.q1:.3f}, 75%={analysis.q3:.3f}, 99%={analysis.percentiles['p99']:.3f}\")\n",
-    "        print(f\"\\n📈 Shape Analysis:\")\n",
+    "        print(\"\\n📈 Shape Analysis:\")\n",
     "        skew_label = '(Right-skewed)' if analysis.skewness > 0.5 else '(Left-skewed)' if analysis.skewness < -0.5 else '(Symmetric)'\n",
     "        print(f\"   Skewness: {analysis.skewness:.2f} {skew_label}\")\n",
     "        kurt_label = '(Heavy tails/outliers)' if analysis.kurtosis > 3 else '(Light tails)'\n",
     "        print(f\"   Kurtosis: {analysis.kurtosis:.2f} {kurt_label}\")\n",
     "        print(f\"   Zeros: {analysis.zero_count:,} ({analysis.zero_percentage:.1f}%)\")\n",
     "        print(f\"   Outliers (IQR): {analysis.outlier_count_iqr:,} ({analysis.outlier_percentage:.1f}%)\")\n",
-    "        \n",
+    "\n",
     "        if rec:\n",
     "            transform_display = TRANSFORM_DISPLAY_NAMES.get(rec.recommended_transform.value, rec.recommended_transform.value)\n",
     "            print(f\"\\n🔧 Recommended Transformation: {transform_display}\")\n",
@@ -1149,20 +1152,20 @@
     "    rec = recommendations.get(col_name)\n",
     "    if not analysis:\n",
     "        continue\n",
-    "    \n",
+    "\n",
     "    data = df[col_name].dropna()\n",
     "    fig = go.Figure()\n",
-    "    \n",
+    "\n",
     "    fig.add_trace(go.Histogram(x=data, nbinsx=50, name='Distribution',\n",
     "                                marker_color='steelblue', opacity=0.7))\n",
-    "    \n",
+    "\n",
     "    mean_val = data.mean()\n",
     "    median_val = data.median()\n",
-    "    \n",
+    "\n",
     "    # Position labels on opposite sides to avoid overlap\n",
     "    mean_position = \"top right\" if mean_val >= median_val else \"top left\"\n",
     "    median_position = \"top left\" if mean_val >= median_val else \"top right\"\n",
-    "    \n",
+    "\n",
     "    fig.add_vline(\n",
     "        x=mean_val, line_dash=\"dash\", line_color=\"red\",\n",
     "        annotation_text=f\"Mean: {mean_val:.2f}\",\n",
@@ -1170,7 +1173,7 @@
     "        annotation_font_color=\"red\",\n",
     "        annotation_bgcolor=\"rgba(255,255,255,0.8)\"\n",
     "    )\n",
-    "    \n",
+    "\n",
     "    fig.add_vline(\n",
     "        x=median_val, line_dash=\"solid\", line_color=\"green\",\n",
     "        annotation_text=f\"Median: {median_val:.2f}\",\n",
@@ -1178,7 +1181,7 @@
     "        annotation_font_color=\"green\",\n",
     "        annotation_bgcolor=\"rgba(255,255,255,0.8)\"\n",
     "    )\n",
-    "    \n",
+    "\n",
     "    # Add 99th percentile marker if there are outliers\n",
     "    if analysis.outlier_percentage > 5:\n",
     "        fig.add_vline(x=analysis.percentiles['p99'], line_dash=\"dot\", line_color=\"orange\",\n",
@@ -1186,7 +1189,7 @@
     "                      annotation_position=\"top right\",\n",
     "                      annotation_font_color=\"orange\",\n",
     "                      annotation_bgcolor=\"rgba(255,255,255,0.8)\")\n",
-    "    \n",
+    "\n",
     "    transform_key = rec.recommended_transform.value if rec else \"none\"\n",
     "    transform_label = TRANSFORM_DISPLAY_NAMES.get(transform_key, transform_key)\n",
     "    fig.update_layout(\n",
@@ -1233,12 +1236,12 @@
     "for col_name in categorical_cols:\n",
     "    col_info = findings.columns[col_name]\n",
     "    cardinality = col_info.universal_metrics.get('distinct_count', df[col_name].nunique())\n",
-    "    \n",
+    "\n",
     "    print(f\"\\n{'='*50}\")\n",
     "    print(f\"Column: {col_name}\")\n",
     "    print(f\"Type: {col_info.inferred_type.value} (Confidence: {col_info.confidence:.0%})\")\n",
     "    print(f\"Distinct Values: {cardinality}\")\n",
-    "    \n",
+    "\n",
     "    # Encoding recommendation based on type and cardinality\n",
     "    if col_info.inferred_type.value == 'categorical_cyclical':\n",
     "        encoding_rec = \"Sin/Cos encoding (cyclical)\"\n",
@@ -1249,7 +1252,7 @@
     "    else:\n",
     "        encoding_rec = \"Target encoding or Frequency encoding (high cardinality)\"\n",
     "    print(f\"Recommended Encoding: {encoding_rec}\")\n",
-    "    \n",
+    "\n",
     "    # Value counts visualization\n",
     "    value_counts = df[col_name].value_counts().head(10)\n",
     "    fig = charts.bar_chart(value_counts.index.tolist(), value_counts.values.tolist(),\n",
@@ -1314,7 +1317,7 @@
     "    # Sort by priority\n",
     "    priority_order = {'high': 0, 'medium': 1, 'low': 2}\n",
     "    transformations.sort(key=lambda x: priority_order.get(x['priority'], 3))\n",
-    "    \n",
+    "\n",
     "    for t in transformations:\n",
     "        priority_marker = \"🔴\" if t['priority'] == 'high' else \"🟡\" if t['priority'] == 'medium' else \"🟢\"\n",
     "        print(f\"\\n   {priority_marker} {t['column']}: {t['transform']}\")\n",
@@ -1352,7 +1355,7 @@
     "    print(\"TEMPORAL AGGREGATION PERSPECTIVE\")\n",
     "    print(\"=\"*70)\n",
     "    print(f\"\\nMedian inter-event time: {median_iet:.0f} days\")\n",
-    "    print(f\"Expected events per window (at median cadence):\")\n",
+    "    print(\"Expected events per window (at median cadence):\")\n",
     "    windows_days = [(\"7d\", 7), (\"30d\", 30), (\"90d\", 90), (\"180d\", 180), (\"365d\", 365)]\n",
     "    for label, days in windows_days:\n",
     "        expected = days / median_iet if median_iet > 0 else 0\n",
@@ -1360,7 +1363,7 @@
     "        print(f\"   {marker} {label}: ~{expected:.1f} events/entity\")\n",
     "\n",
     "    # Within-entity vs between-entity variance per column\n",
-    "    print(f\"\\nColumn Temporal Variability (within-entity CV vs between-entity CV):\")\n",
+    "    print(\"\\nColumn Temporal Variability (within-entity CV vs between-entity CV):\")\n",
     "    print(f\"{'Column':<25} {'Within-CV':<12} {'Between-CV':<12} {'Ratio':<8} {'Aggregation Guidance'}\")\n",
     "    print(\"-\" * 90)\n",
     "\n",
@@ -1390,11 +1393,11 @@
     "        ratio_str = f\"{ratio:.2f}\" if not np.isinf(ratio) else \">10\"\n",
     "        print(f\"{col:<25} {within_str:<12} {between_cv:<12.2f} {ratio_str:<8} {guidance}\")\n",
     "\n",
-    "    print(f\"\\nInterpretation:\")\n",
-    "    print(f\"  Within-CV: how much each entity\\'s values vary across their events\")\n",
-    "    print(f\"  Between-CV: how much entity averages differ from each other\")\n",
-    "    print(f\"  Ratio > 1: temporal variation dominates -> shorter windows capture dynamics\")\n",
-    "    print(f\"  Ratio < 1: entity identity dominates -> longer windows (or all_time) sufficient\")\n"
+    "    print(\"\\nInterpretation:\")\n",
+    "    print(\"  Within-CV: how much each entity\\'s values vary across their events\")\n",
+    "    print(\"  Between-CV: how much entity averages differ from each other\")\n",
+    "    print(\"  Ratio > 1: temporal variation dominates -> shorter windows capture dynamics\")\n",
+    "    print(\"  Ratio < 1: entity identity dominates -> longer windows (or all_time) sufficient\")\n"
    ]
   },
   {
@@ -1466,8 +1469,8 @@
     "explanation[\"meaningful_pct\"] = (explanation[\"meaningful_pct\"] * 100).round(1).astype(str) + \"%\"\n",
     "display_table(explanation)\n",
     "\n",
-    "print(f\"\\nCoverage: % of entities with enough tenure AND expected >=2 events in that window\")\n",
-    "print(f\"Meaningful: among entities with enough tenure, % that have sufficient event density\")"
+    "print(\"\\nCoverage: % of entities with enough tenure AND expected >=2 events in that window\")\n",
+    "print(\"Meaningful: among entities with enough tenure, % that have sufficient event density\")"
    ]
   },
   {
@@ -1495,9 +1498,9 @@
     "h = window_result.heterogeneity\n",
     "\n",
     "print(\"Temporal Heterogeneity (eta-squared):\")\n",
-    "print(f\"  eta² measures the fraction of variance in a metric explained by lifecycle quadrant grouping.\")\n",
-    "print(f\"  Scale: 0 = no group differences, 1 = all variance is between groups.\")\n",
-    "print(f\"  Thresholds: <0.06 = low | 0.06-0.14 = moderate | >0.14 = high effect size\\n\")\n",
+    "print(\"  eta² measures the fraction of variance in a metric explained by lifecycle quadrant grouping.\")\n",
+    "print(\"  Scale: 0 = no group differences, 1 = all variance is between groups.\")\n",
+    "print(\"  Thresholds: <0.06 = low | 0.06-0.14 = moderate | >0.14 = high effect size\\n\")\n",
     "\n",
     "eta_max = max(h.eta_squared_intensity, h.eta_squared_event_count)\n",
     "print(f\"  Intensity eta²:   {h.eta_squared_intensity:.3f}  {'<-- dominant' if h.eta_squared_intensity >= h.eta_squared_event_count else ''}\")\n",

{churnkit-0.76.0a3.data → churnkit-0.76.1a2.data}/data/share/churnkit/exploration_notebooks/01b_temporal_quality.ipynb

@@ -71,22 +71,25 @@
    "outputs": [],
    "source": [
     "from customer_retention.analysis.notebook_progress import track_and_export_previous\n",
+    "\n",
     "track_and_export_previous(\"01b_temporal_quality.ipynb\")\n",
     "\n",
-    "from pathlib import Path\n",
     "import pandas as pd\n",
     "import plotly.graph_objects as go\n",
-    "from plotly.subplots import make_subplots\n",
     "\n",
     "from customer_retention.analysis.auto_explorer import ExplorationFindings, RecommendationEngine\n",
     "from customer_retention.analysis.visualization import ChartBuilder, display_figure\n",
     "from customer_retention.core.config.column_config import ColumnType\n",
+    "from customer_retention.core.config.experiments import FINDINGS_DIR\n",
     "from customer_retention.stages.profiling import (\n",
-    "    DuplicateEventCheck, TemporalGapCheck, FutureDateCheck, EventOrderCheck,\n",
-    "    TemporalQualityReporter, SegmentAwareOutlierAnalyzer\n",
+    "    DuplicateEventCheck,\n",
+    "    EventOrderCheck,\n",
+    "    FutureDateCheck,\n",
+    "    SegmentAwareOutlierAnalyzer,\n",
+    "    TemporalGapCheck,\n",
+    "    TemporalQualityReporter,\n",
     ")\n",
-    "from customer_retention.stages.temporal import load_data_with_snapshot_preference, TEMPORAL_METADATA_COLS\n",
-    "from customer_retention.core.config.experiments import FINDINGS_DIR, EXPERIMENTS_DIR, OUTPUT_DIR, setup_experiments_structure\n"
+    "from customer_retention.stages.temporal import TEMPORAL_METADATA_COLS, load_data_with_snapshot_preference\n"
    ]
   },
   {
@@ -397,7 +400,7 @@
     "if numeric_cols:\n",
     "    analyzer = SegmentAwareOutlierAnalyzer(max_segments=5)\n",
     "    result = analyzer.analyze(df, feature_cols=numeric_cols, segment_col=None, target_col=findings.target_column)\n",
-    "    \n",
+    "\n",
     "    print(f\"Segments detected: {result.n_segments}\")\n",
     "    if result.n_segments > 1:\n",
     "        data = [{\"Feature\": c, \"Global\": result.global_analysis[c].outliers_detected,\n",
@@ -456,25 +459,7 @@
     "tags": []
    },
    "outputs": [],
-   "source": [
-    "# Binary field validation\n",
-    "binary_cols = [n for n, c in findings.columns.items() if c.inferred_type == ColumnType.BINARY and n not in TEMPORAL_METADATA_COLS]\n",
-    "for col in binary_cols:\n",
-    "    c0, c1 = (df[col] == 0).sum(), (df[col] == 1).sum()\n",
-    "    print(f\"✓ {col}: 0={c0:,} ({c0/(c0+c1)*100:.1f}%), 1={c1:,} ({c1/(c0+c1)*100:.1f}%)\")\n",
-    "\n",
-    "# Consistency check\n",
-    "issues = []\n",
-    "for col in df.select_dtypes(include=['object']).columns:\n",
-    "    if col in [ENTITY_COLUMN, TIME_COLUMN]: continue\n",
-    "    variants = {}\n",
-    "    for v in df[col].dropna().unique():\n",
-    "        key = str(v).lower().strip()\n",
-    "        variants.setdefault(key, []).append(v)\n",
-    "    issues.extend([{\"Column\": col, \"Variants\": vs} for vs in variants.values() if len(vs) > 1])\n",
-    "\n",
-    "print(f\"\\n{'⚠️ Consistency issues: ' + str(len(issues)) if issues else '✅ No consistency issues'}\")"
-   ]
+   "source": "# Binary field validation\nbinary_cols = [n for n, c in findings.columns.items() if c.inferred_type == ColumnType.BINARY and n not in TEMPORAL_METADATA_COLS]\nfor col in binary_cols:\n    c0, c1 = (df[col] == 0).sum(), (df[col] == 1).sum()\n    print(f\"✓ {col}: 0={c0:,} ({c0/(c0+c1)*100:.1f}%), 1={c1:,} ({c1/(c0+c1)*100:.1f}%)\")\n\n# Consistency check\nissues = []\nfor col in df.select_dtypes(include=['object']).columns:\n    if col in [ENTITY_COLUMN, TIME_COLUMN]:\n        continue\n    variants = {}\n    for v in df[col].dropna().unique():\n        key = str(v).lower().strip()\n        variants.setdefault(key, []).append(v)\n    issues.extend([{\"Column\": col, \"Variants\": vs} for vs in variants.values() if len(vs) > 1])\n\nprint(f\"\\n{'⚠️ Consistency issues: ' + str(len(issues)) if issues else '✅ No consistency issues'}\")"
   },
   {
    "cell_type": "markdown",
@@ -676,4 +661,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 5
-}
+}