Create appv2.py

appv2.py

@@ -0,0 +1,823 @@
+# app.py
+
+import pandas as pd
+import gradio as gr
+
+# ---------------------------------------------------
+# HELPERS
+# ---------------------------------------------------
+
+from helper.vintage_helpers import (
+    create_booking_vintage
+)
+
+from helper.data_merger import (
+    merge_acq_perf
+)
+
+# ---------------------------------------------------
+# METRICS
+# ---------------------------------------------------
+
+from metrics.mix_metrics import (
+    calculate_vintage_mix,
+    calculate_limit_mix
+)
+
+# ---------------------------------------------------
+# ANALYTICS
+# ---------------------------------------------------
+
+from analytics.performance_analysis import (
+    generate_metric_view
+)
+
+# ---------------------------------------------------
+# VISUALIZATIONS - VINTAGE CURVES
+# ---------------------------------------------------
+
+from visualizations.vintage_curves import (
+    generate_delinquency_metric_chart,
+    generate_multi_metric_comparison,
+    generate_segment_delinquency_curve
+)
+
+# ---------------------------------------------------
+# VISUALIZATIONS - SEGMENT RANKING
+# ---------------------------------------------------
+
+from visualizations.segment_ranking import (
+    generate_segment_risk_heatmap,
+    generate_segment_risk_ranking,
+    generate_multi_category_risk_comparison,
+    calculate_portfolio_risk_summary
+)
+
+# ---------------------------------------------------
+# LOAD DATA
+# ---------------------------------------------------
+
+acq = pd.read_csv(
+    "data/acquisition.csv"
+)
+
+perf = pd.read_csv(
+    "data/performance.csv"
+)
+
+# ---------------------------------------------------
+# CREATE BOOKING VINTAGE
+# ---------------------------------------------------
+
+acq = create_booking_vintage(
+    acq,
+    booking_date_col="booking_date"
+)
+
+# ---------------------------------------------------
+# CREATE MASTER PERFORMANCE DATASET
+# ---------------------------------------------------
+
+master_df = merge_acq_perf(
+    acq_df=acq,
+    perf_df=perf
+)
+
+# ---------------------------------------------------
+# ACQUISITION ANALYSIS
+# ---------------------------------------------------
+
+def run_acquisition_analysis(
+    analysis_type,
+    category
+):
+
+    # -----------------------------------------
+    # PORTFOLIO MIX
+    # -----------------------------------------
+
+    if analysis_type == "Portfolio Mix":
+
+        result = (
+            acq.groupby(
+                ["booking_vintage", category]
+            )
+            .agg(
+                count=("account_id", "nunique"),
+                balance=("credit_limit", "sum")
+            )
+            .reset_index()
+        )
+
+        vintage_total = (
+            result.groupby("booking_vintage")["count"]
+            .transform("sum")
+        )
+
+        result["rate"] = (
+            result["count"] / vintage_total
+        ) * 100
+
+        result["rate"] = (
+            result["rate"]
+            .round(2)
+        )
+
+    # -----------------------------------------
+    # CREDIT LINE CONCENTRATION
+    # -----------------------------------------
+
+    elif analysis_type == "Credit Line Concentration":
+
+        result = (
+            acq.groupby(
+                ["booking_vintage", category]
+            )
+            .agg(
+                count=("account_id", "nunique"),
+                balance=("credit_limit", "sum")
+            )
+            .reset_index()
+        )
+
+        vintage_total = (
+            result.groupby("booking_vintage")["balance"]
+            .transform("sum")
+        )
+
+        result["rate"] = (
+            result["balance"] / vintage_total
+        ) * 100
+
+        result["rate"] = (
+            result["rate"]
+            .round(2)
+        )
+
+    else:
+
+        return pd.DataFrame()
+
+    # -----------------------------------------
+    # STANDARDIZED OUTPUT
+    # -----------------------------------------
+
+    result = result.rename(
+        columns={
+            "booking_vintage": "Vintage",
+            category: "Category",
+            "count": "Count",
+            "balance": "Balance",
+            "rate": "Rate"
+        }
+    )
+
+    return result[
+        [
+            "Vintage",
+            "Category",
+            "Count",
+            "Balance",
+            "Rate"
+        ]
+    ]
+
+# ---------------------------------------------------
+# PERFORMANCE ANALYSIS
+# ---------------------------------------------------
+
+def run_performance_analysis(
+    metric_name,
+    view_level
+):
+
+    # -----------------------------------------
+    # VIEW MAPPING
+    # -----------------------------------------
+
+    view_mapping = {
+
+        "Overall": None,
+
+        "Channel":
+        "sourcing_channel",
+
+        "FICO":
+        "fico_band",
+
+        "City Tier":
+        "city_tier",
+
+        "Occupation":
+        "occupation_type"
+    }
+
+    group_col = view_mapping[
+        view_level
+    ]
+
+    # -----------------------------------------
+    # CALL ANALYTICS ENGINE
+    # -----------------------------------------
+
+    result = generate_metric_view(
+        df=master_df,
+        metric_name=metric_name,
+        group_col=group_col
+    )
+
+    # -----------------------------------------
+    # STANDARDIZE OUTPUT
+    # -----------------------------------------
+
+    if group_col is not None:
+
+        result = result.rename(
+            columns={
+                group_col: "Category"
+            }
+        )
+
+    else:
+
+        result["Category"] = "Overall"
+
+    # -----------------------------------------
+    # IDENTIFY RATE COLUMN
+    # -----------------------------------------
+
+    rate_col = [
+        col for col in result.columns
+        if "rate" in col.lower()
+    ][0]
+
+    # -----------------------------------------
+    # OUTPUT FORMAT
+    # -----------------------------------------
+
+    final_result = pd.DataFrame()
+
+    final_result["Vintage"] = (
+        result["booking_vintage"]
+    )
+
+    final_result["Category"] = (
+        result["Category"]
+    )
+
+    final_result["Count"] = (
+        result["total_accounts"]
+    )
+
+    final_result["Balance"] = (
+        result["total_balance"]
+    )
+
+    final_result["Rate"] = (
+        result[rate_col]
+        .round(2)
+    )
+
+    return final_result
+
+# ---------------------------------------------------
+# VINTAGE CURVES ANALYSIS
+# ---------------------------------------------------
+
+def generate_vintage_curve_single(
+    metric_name
+):
+    """Generate single vintage curve for a metric."""
+    try:
+        fig = generate_delinquency_metric_chart(
+            df=master_df,
+            metric_name=metric_name,
+            chart_type="line"
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating vintage curve: {str(e)}"
+
+
+def generate_vintage_curves_comparison():
+    """Generate comparison of all vintage curves."""
+    try:
+        fig = generate_multi_metric_comparison(
+            df=master_df,
+            metrics=["30+@3", "30+@6", "60+@6", "Yr1 NCL"]
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating comparison: {str(e)}"
+
+
+def generate_segmented_vintage_curve(
+    metric_name,
+    category
+):
+    """Generate vintage curve segmented by category."""
+    try:
+        fig = generate_segment_delinquency_curve(
+            df=master_df,
+            metric_name=metric_name,
+            category=category
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating segmented curve: {str(e)}"
+
+# ---------------------------------------------------
+# SEGMENT RANKING ANALYSIS
+# ---------------------------------------------------
+
+def generate_segment_risk_heatmap_chart():
+    """Generate risk heatmap across all segments and metrics."""
+    try:
+        fig = generate_segment_risk_heatmap(
+            df=master_df
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating heatmap: {str(e)}"
+
+
+def generate_high_risk_segments_ranking(
+    metric_name,
+    category
+):
+    """Generate ranking of high-risk segments."""
+    try:
+        fig = generate_segment_risk_ranking(
+            df=master_df,
+            metric_name=metric_name,
+            category=category,
+            top_n=10
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating ranking: {str(e)}"
+
+
+def generate_multi_category_comparison(
+    metric_name
+):
+    """Generate risk comparison across all categories."""
+    try:
+        fig = generate_multi_category_risk_comparison(
+            df=master_df,
+            metric_name=metric_name
+        )
+        return fig
+    except Exception as e:
+        return f"Error generating comparison: {str(e)}"
+
+
+def generate_portfolio_summary():
+    """Generate portfolio risk summary."""
+    try:
+        summary_df = calculate_portfolio_risk_summary(
+            df=master_df
+        )
+        return summary_df
+    except Exception as e:
+        return f"Error generating summary: {str(e)}"
+
+# ---------------------------------------------------
+# DYNAMIC DROPDOWNS
+# ---------------------------------------------------
+
+def update_analysis_dropdown(
+    dataset
+):
+
+    # -----------------------------------------
+    # ACQUISITION
+    # -----------------------------------------
+
+    if dataset == "Acquisition":
+
+        return gr.update(
+            choices=[
+                "Portfolio Mix",
+                "Credit Line Concentration"
+            ],
+            value="Portfolio Mix"
+        )
+
+    # -----------------------------------------
+    # PERFORMANCE
+    # -----------------------------------------
+
+    elif dataset == "Performance":
+
+        return gr.update(
+            choices=[
+                "30+@3",
+                "30+@6",
+                "60+@6",
+                "Yr1 NCL"
+            ],
+            value="30+@6"
+        )
+
+
+def update_category_dropdown(
+    dataset
+):
+
+    # -----------------------------------------
+    # ACQUISITION
+    # -----------------------------------------
+
+    if dataset == "Acquisition":
+
+        return gr.update(
+            choices=[
+                "fico_band",
+                "sourcing_channel",
+                "city_tier",
+                "occupation_type"
+            ],
+            value="fico_band"
+        )
+
+    # -----------------------------------------
+    # PERFORMANCE
+    # -----------------------------------------
+
+    elif dataset == "Performance":
+
+        return gr.update(
+            choices=[
+                "Overall",
+                "Channel",
+                "FICO",
+                "City Tier",
+                "Occupation"
+            ],
+            value="Overall"
+        )
+
+# ---------------------------------------------------
+# MASTER ROUTER
+# ---------------------------------------------------
+
+def run_analysis(
+    dataset,
+    analysis,
+    category
+):
+
+    # -----------------------------------------
+    # ACQUISITION
+    # -----------------------------------------
+
+    if dataset == "Acquisition":
+
+        return run_acquisition_analysis(
+            analysis_type=analysis,
+            category=category
+        )
+
+    # -----------------------------------------
+    # PERFORMANCE
+    # -----------------------------------------
+
+    elif dataset == "Performance":
+
+        return run_performance_analysis(
+            metric_name=analysis,
+            view_level=category
+        )
+
+    else:
+
+        return pd.DataFrame()
+
+# ---------------------------------------------------
+# GRADIO UI
+# ---------------------------------------------------
+
+with gr.Blocks() as app:
+
+    gr.Markdown(
+        "# Risk Analytics Manager Agent - Phase 2"
+    )
+
+    with gr.Tabs():
+
+        # =================================================
+        # TAB 1: BASIC ANALYSIS (Phase 1)
+        # =================================================
+
+        with gr.TabItem("📊 Basic Analysis"):
+
+            gr.Markdown(
+                "## Phase 1: Acquisition & Performance Analysis"
+            )
+
+            with gr.Row():
+
+                dataset_dropdown = gr.Dropdown(
+                    choices=[
+                        "Acquisition",
+                        "Performance"
+                    ],
+                    value="Acquisition",
+                    label="Dataset"
+                )
+
+                analysis_dropdown = gr.Dropdown(
+                    choices=[
+                        "Portfolio Mix",
+                        "Credit Line Concentration"
+                    ],
+                    value="Portfolio Mix",
+                    label="Analysis"
+                )
+
+                category_dropdown = gr.Dropdown(
+                    choices=[
+                        "fico_band",
+                        "sourcing_channel",
+                        "city_tier",
+                        "occupation_type"
+                    ],
+                    value="fico_band",
+                    label="Category / View"
+                )
+
+            # -----------------------------------------
+            # DYNAMIC DROPDOWNS
+            # -----------------------------------------
+
+            dataset_dropdown.change(
+                fn=update_analysis_dropdown,
+                inputs=dataset_dropdown,
+                outputs=analysis_dropdown
+            )
+
+            dataset_dropdown.change(
+                fn=update_category_dropdown,
+                inputs=dataset_dropdown,
+                outputs=category_dropdown
+            )
+
+            # -----------------------------------------
+            # RUN BUTTON
+            # -----------------------------------------
+
+            run_button = gr.Button(
+                "Run Analysis",
+                variant="primary"
+            )
+
+            output_table = gr.Dataframe()
+
+            run_button.click(
+                fn=run_analysis,
+                inputs=[
+                    dataset_dropdown,
+                    analysis_dropdown,
+                    category_dropdown
+                ],
+                outputs=output_table
+            )
+
+        # =================================================
+        # TAB 2: VINTAGE CURVES (Phase 2)
+        # =================================================
+
+        with gr.TabItem("📈 Vintage Curves"):
+
+            gr.Markdown(
+                "## Phase 2: Vintage Delinquency Curves Analysis"
+            )
+
+            with gr.Row():
+
+                metric_dropdown = gr.Dropdown(
+                    choices=[
+                        "30+@3",
+                        "30+@6",
+                        "60+@6",
+                        "Yr1 NCL"
+                    ],
+                    value="30+@6",
+                    label="Delinquency Metric"
+                )
+
+                vintage_chart_type = gr.Radio(
+                    choices=["Single Metric", "All Metrics Comparison"],
+                    value="Single Metric",
+                    label="Chart Type"
+                )
+
+            def update_vintage_view(metric, chart_type):
+                if chart_type == "Single Metric":
+                    return generate_vintage_curve_single(metric)
+                else:
+                    return generate_vintage_curves_comparison()
+
+            vintage_chart = gr.Plot(
+                label="Vintage Curve"
+            )
+
+            gen_vintage_btn = gr.Button(
+                "Generate Vintage Curve",
+                variant="primary"
+            )
+
+            gen_vintage_btn.click(
+                fn=update_vintage_view,
+                inputs=[metric_dropdown, vintage_chart_type],
+                outputs=vintage_chart
+            )
+
+            gr.Markdown(
+                "### Segmented Vintage Curves"
+            )
+
+            with gr.Row():
+
+                segment_metric = gr.Dropdown(
+                    choices=[
+                        "30+@3",
+                        "30+@6",
+                        "60+@6",
+                        "Yr1 NCL"
+                    ],
+                    value="30+@6",
+                    label="Metric"
+                )
+
+                segment_category = gr.Dropdown(
+                    choices=[
+                        "fico_band",
+                        "sourcing_channel",
+                        "city_tier",
+                        "occupation_type"
+                    ],
+                    value="fico_band",
+                    label="Category"
+                )
+
+            segmented_chart = gr.Plot(
+                label="Segmented Vintage Curve"
+            )
+
+            gen_segment_btn = gr.Button(
+                "Generate Segmented Curve",
+                variant="primary"
+            )
+
+            gen_segment_btn.click(
+                fn=generate_segmented_vintage_curve,
+                inputs=[segment_metric, segment_category],
+                outputs=segmented_chart
+            )
+
+        # =================================================
+        # TAB 3: SEGMENT RANKING (Phase 2)
+        # =================================================
+
+        with gr.TabItem("⚠️ Segment Ranking"):
+
+            gr.Markdown(
+                "## Phase 2: High-Risk Segment Analysis"
+            )
+
+            # --------- HEATMAP SECTION ---------
+
+            gr.Markdown(
+                "### 🔥 Overall Risk Heatmap"
+            )
+
+            gr.Markdown(
+                "Risk scores across all delinquency metrics and segments"
+            )
+
+            heatmap_chart = gr.Plot(
+                label="Risk Heatmap"
+            )
+
+            gen_heatmap_btn = gr.Button(
+                "Generate Risk Heatmap",
+                variant="primary"
+            )
+
+            gen_heatmap_btn.click(
+                fn=generate_segment_risk_heatmap_chart,
+                outputs=heatmap_chart
+            )
+
+            gr.Markdown(
+                "---"
+            )
+
+            # --------- HIGH-RISK RANKING SECTION ---------
+
+            gr.Markdown(
+                "### 📊 High-Risk Segments Ranking"
+            )
+
+            with gr.Row():
+
+                ranking_metric = gr.Dropdown(
+                    choices=[
+                        "30+@3",
+                        "30+@6",
+                        "60+@6",
+                        "Yr1 NCL"
+                    ],
+                    value="30+@6",
+                    label="Metric"
+                )
+
+                ranking_category = gr.Dropdown(
+                    choices=[
+                        "fico_band",
+                        "sourcing_channel",
+                        "city_tier",
+                        "occupation_type"
+                    ],
+                    value="fico_band",
+                    label="Category"
+                )
+
+            ranking_chart = gr.Plot(
+                label="High-Risk Segments"
+            )
+
+            gen_ranking_btn = gr.Button(
+                "Generate Risk Ranking",
+                variant="primary"
+            )
+
+            gen_ranking_btn.click(
+                fn=generate_high_risk_segments_ranking,
+                inputs=[ranking_metric, ranking_category],
+                outputs=ranking_chart
+            )
+
+            gr.Markdown(
+                "---"
+            )
+
+            # --------- MULTI-CATEGORY COMPARISON ---------
+
+            gr.Markdown(
+                "### 🔀 Cross-Category Risk Comparison"
+            )
+
+            comparison_metric = gr.Dropdown(
+                choices=[
+                    "30+@3",
+                    "30+@6",
+                    "60+@6",
+                    "Yr1 NCL"
+                ],
+                value="30+@6",
+                label="Metric"
+            )
+
+            comparison_chart = gr.Plot(
+                label="Multi-Category Comparison"
+            )
+
+            gen_comparison_btn = gr.Button(
+                "Generate Comparison",
+                variant="primary"
+            )
+
+            gen_comparison_btn.click(
+                fn=generate_multi_category_comparison,
+                inputs=comparison_metric,
+                outputs=comparison_chart
+            )
+
+            gr.Markdown(
+                "---"
+            )
+
+            # --------- PORTFOLIO SUMMARY ---------
+
+            gr.Markdown(
+                "### 📋 Portfolio Risk Summary"
+            )
+
+            summary_table = gr.Dataframe(
+                label="Risk Summary"
+            )
+
+            gen_summary_btn = gr.Button(
+                "Generate Summary",
+                variant="primary"
+            )
+
+            gen_summary_btn.click(
+                fn=generate_portfolio_summary,
+                outputs=summary_table
+            )
+
+app.launch()