Sync from GitHub via hub-sync

README.md

@@ -8,20 +8,21 @@ app_file: app.py
 pinned: false
 ---
 
-# DAIOE Years Explorer — Swedish Occupations
+# DAIOE Years Explorer: Swedish Occupations
 
 An interactive Shiny app for exploring AI exposure and employment trends across Swedish occupations, built on the [DAIOE](https://www.ai-econlab.com/ai-exposure-daioe) index linked to [SCB's Swedish Occupational Register](https://www.scb.se/en/finding-statistics/statistics-by-subject-area/labour-market/labour-force-supply/the-swedish-occupational-register-with-statistics/).
 
 ## Features
 
-- **Occupation View** — Select an occupation by SSYK level and year to see employment statistics and % changes (1, 3, 5-year)
-- **AI Exposure** — Ranked bar chart of AI sub-domain exposure scores with percentile rankings and exposure levels
-- **Employment by Age Group** — Line chart of annual employment % change by age group over a custom year range
-- **Download** — Filter and export the full dataset as CSV, Parquet, or Excel
+- **Occupation View**: Select an occupation by SSYK level and year to see employment statistics and % changes (1, 3, 5-year)
+- **AI Exposure**: Ranked bar chart of AI sub-domain exposure scores with percentile rankings and exposure levels
+- **Employment by Age Group**: Line chart of annual employment % change by age group over a custom year range
+- **Comparison View**: Benchmark multiple occupations side by side with an employment trend chart, summary table, and AI exposure radar chart
+- **Download**: Filter and export the full dataset as CSV, Parquet, or Excel
 
 ## Data Sources
 
 | Data | Source |
 |------|--------|
-| AI Exposure Index | [DAIOE — AI Econ Lab](https://www.ai-econlab.com/ai-exposure-daioe) |
-| Employment Statistics | [Swedish Occupational Register, SCB](https://www.scb.se/en/finding-statistics/statistics-by-subject-area/labour-market/labour-force-supply/the-swedish-occupational-register-with-statistics/)
+| AI Exposure Index | [DAIOE - AI Econ Lab](https://www.ai-econlab.com/ai-exposure-daioe) |
+| Employment Statistics | [Swedish Occupational Register, SCB](https://www.scb.se/en/finding-statistics/statistics-by-subject-area/labour-market/labour-force-supply/the-swedish-occupational-register-with-statistics/) |

app.py

@@ -19,6 +19,7 @@ from src.setup import (
     build_choices_by_level,
     download_extension,
     download_media_type,
+    empty_figure,
     export_filtered_data,
     lf,
 )
@@ -76,6 +77,25 @@ def occ_summary():
 
 
 
+@reactive.calc
+def comparison_data():
+    """Return total employment per year/occupation for the comparison view."""
+    occs = list(input.comp_occs())
+    ages = list(input.comp_age())
+    if not occs or not ages:
+        return pl.DataFrame()
+    return calcs.get_comparison_employment(lf, occs, ages)
+
+
+@reactive.calc
+def comp_radar_data():
+    """Return mean AI percentile scores per occupation for the radar chart."""
+    occs = list(input.comp_occs())
+    if not occs:
+        return pl.DataFrame()
+    return calcs.get_comp_radar(lf, occs, int(input.comp_year()))
+
+
 @reactive.calc
 def occ_employment_by_age():
     """Reactive wrapper: returns long-format employment by age group for the line chart."""
@@ -168,7 +188,79 @@ with ui.navset_pill(id="tab"):
                 "Card 4"
 
     with ui.nav_panel(title="2. Comparison View"):
-        "Panel B content"
+        with ui.layout_sidebar():
+            with ui.sidebar(bg="#FFFFFF", width=300, title="Benchmarking"):
+                ui.input_select("comp_level", "SSYK Level", choices=["All Levels", *LEVELS], selected=DEFAULT_LEVEL)
+                ui.input_selectize("comp_occs", "Select Occupations", choices={}, multiple=True)
+                ui.hr()
+                ui.input_selectize("comp_age", "Age Group", choices=AGES, selected="Early Career 2 (25-29)", multiple=True)
+                ui.hr()
+                ui.input_select("comp_year", "Comparison Year (Radar)", choices=[str(y) for y in YEARS], selected=str(YEAR_MAX))
+
+            with ui.card():
+                ui.card_header("Benchmarking Summary")
+
+                @render.ui
+                def comparison_summary():
+                    df = comparison_data()
+                    if df.is_empty():
+                        return ui.markdown("*Select occupations to generate a summary...*")
+
+                    latest_yr = df["year"].max()
+                    summary_rows = []
+                    for occ in df["occupation"].unique():
+                        sub = df.filter(pl.col("occupation") == occ).sort("year")
+                        curr_emp = sub.tail(1)["count"][0]
+
+                        def _val(yr, _sub=sub):
+                            s = _sub.filter(pl.col("year") == yr)["count"]
+                            return f"{int(s[0]):,}" if not s.is_empty() else "---"
+
+                        summary_rows.append(
+                            ui.tags.tr(
+                                ui.tags.td(occ, style="font-weight: bold;"),
+                                ui.tags.td(_val(latest_yr - 5)),
+                                ui.tags.td(_val(latest_yr - 3)),
+                                ui.tags.td(_val(latest_yr - 1)),
+                                ui.tags.td(f"{int(curr_emp):,}", style="background-color: #f8f9fa; font-weight: bold;"),
+                            ),
+                        )
+
+                    return ui.tags.table(
+                        ui.tags.thead(
+                            ui.tags.tr(
+                                ui.tags.th("Occupation"),
+                                ui.tags.th(f"Emp ({latest_yr - 5})"),
+                                ui.tags.th(f"Emp ({latest_yr - 3})"),
+                                ui.tags.th(f"Emp ({latest_yr - 1})"),
+                                ui.tags.th(f"Emp ({latest_yr})"),
+                            ),
+                        ),
+                        ui.tags.tbody(*summary_rows),
+                        class_="table table-sm table-hover",
+                        style="font-size: 0.9rem;",
+                    )
+
+            with ui.layout_columns(col_widths=[6, 6], gap="1rem"):
+                with ui.card(full_screen=True):
+                    ui.card_header("Employment Trends (Selected Occupations)")
+
+                    @render_plotly
+                    def comparison_employment_plot():
+                        df = comparison_data()
+                        if df.is_empty():
+                            return empty_figure("Select occupations to compare", {"text": "#666"})
+                        return visuals.build_comparison_employment_plot(df.to_pandas())
+
+                with ui.card(full_screen=True):
+                    ui.card_header("Radar Comparison (AI Exposure Percentiles)")
+
+                    @render_plotly
+                    def comp_radar_plot():
+                        df = comp_radar_data()
+                        if df.is_empty():
+                            return empty_figure("Select occupations to compare", {"text": "#666"})
+                        return visuals.build_comp_radar_plot(df.to_pandas(), METRICS)
 
     with ui.nav_panel(title="3. Download"):
         ui.p(
@@ -273,6 +365,17 @@ def _sync_occupation_choices():
     ui.update_selectize("occupation", choices=choices, selected=next(iter(choices)))
 
 
+@reactive.effect
+def _sync_comp_occupation_choices():
+    """Update comparison occupation choices when the SSYK level changes."""
+    level = input.comp_level()
+    if level == "All Levels":
+        choices = {occ: occ for d in OCCUPATION_CHOICES.values() for occ in d}
+    else:
+        choices = OCCUPATION_CHOICES.get(level, {})
+    ui.update_selectize("comp_occs", choices=choices, selected=[])
+
+
 @reactive.effect
 def _sync_download_occupation_choices():
     """Update the download occupation selectize when the download SSYK level changes."""

src/calcs.py

@@ -122,6 +122,50 @@ def get_occ_ai_trend(
     )
 
 
+def get_comparison_employment(
+    lf: pl.LazyFrame,
+    occupations: list[str],
+    age_groups: list[str],
+) -> pl.DataFrame:
+    """
+    Return total employment per year/occupation for the comparison view.
+
+    Aggregates across all sexes and the selected age groups.
+    Returns a DataFrame with columns: year, occupation, count.
+    """
+    return (
+        lf.filter(
+            pl.col("occupation").is_in(occupations)
+            & pl.col("age_group").is_in(age_groups),
+        )
+        .group_by(["year", "occupation"])
+        .agg(pl.col("count").sum())
+        .sort(["occupation", "year"])
+        .collect()
+    )
+
+
+def get_comp_radar(
+    lf: pl.LazyFrame,
+    occupations: list[str],
+    year: int,
+) -> pl.DataFrame:
+    """
+    Return mean AI percentile scores per occupation for the radar chart.
+
+    Returns a DataFrame with columns: occupation, pctl_<metric>_wavg for each metric.
+    """
+    return (
+        lf.filter(
+            pl.col("occupation").is_in(occupations)
+            & (pl.col("year") == year),
+        )
+        .group_by("occupation")
+        .agg([pl.col(c).mean() for c in AI_PCTL_COLS])
+        .collect()
+    )
+
+
 def get_occ_employment_by_age(
     lf: pl.LazyFrame,
     occupation: str,

src/visuals.py

@@ -137,6 +137,58 @@ def build_age_chart(df: pd.DataFrame, occupation: str) -> go.Figure:
     return fig
 
 
+def build_comparison_employment_plot(df: pd.DataFrame) -> go.Figure:
+    """Build a line chart comparing employment trends across selected occupations."""
+    if df.empty:
+        return go.Figure()
+
+    fig = px.line(
+        df,
+        x="year",
+        y="count",
+        color="occupation",
+        markers=True,
+        labels={"count": "Total Employment", "year": "Year"},
+    )
+    fig.update_layout(
+        **_BASE_LAYOUT,
+        legend={"orientation": "h", "yanchor": "bottom", "y": -0.25, "xanchor": "center", "x": 0.5, "title": None},
+    )
+    fig.update_xaxes(gridcolor=_C_GRID, zeroline=False, dtick=1)
+    fig.update_yaxes(gridcolor=_C_GRID, zeroline=False)
+    return fig
+
+
+def build_comp_radar_plot(df: pd.DataFrame, metrics: dict[str, str]) -> go.Figure:
+    """Build a radar chart comparing AI percentile scores across selected occupations."""
+    if df.empty:
+        return go.Figure()
+
+    categories = list(metrics.values())
+    fig = go.Figure()
+
+    for _, row in df.iterrows():
+        r_values = [row[f"pctl_{k}_wavg"] for k in metrics]
+        r_values_closed = [*r_values, r_values[0]]
+        categories_closed = [*categories, categories[0]]
+
+        fig.add_trace(go.Scatterpolar(
+            r=r_values_closed,
+            theta=categories_closed,
+            fill="toself",
+            name=row["occupation"],
+            hovertemplate="%{theta}: %{r:.1f}%<extra></extra>",
+        ))
+
+    fig.update_layout(
+        **_BASE_LAYOUT,
+        polar={"radialaxis": {"visible": True, "range": [0, 100]}},
+        showlegend=True,
+        legend={"orientation": "h", "yanchor": "bottom", "y": -0.25, "xanchor": "center", "x": 0.5},
+    )
+    return fig
+
+
 def build_ai_exposure_bar(df: pd.DataFrame, occupation: str, year: int) -> go.Figure:
     """
     Build a vertical bar chart of AI exposure level per sub-domain.