Drop zoom view, remove past-Toto overlay, dated x-axis

- app.py: remove VIEW_ZOOM and zoom_plot. Only the weekly view renders.
- app.py + src/weather_ui.py: stop building / passing past_toto into
combined_figure. The 'past forecasts' overlay used MAX(forecast_made_at)
per target hour, which sampled forecasts at wildly different lags
depending on autorefresh timing — that's where the sawtooth came from,
not from Toto. The scoreboard MAE is the honest single number; the
chart overlay was the wrong abstraction. (A fixed-horizon overlay is
a future option.)
- src/weather_ui.py: explicit x-axis tick format 'Month day\nhour AM/PM'
so dates are unambiguous on a 7-day-plus-72h chart.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

app.py

@@ -30,18 +30,7 @@ CACHE_TTL_SECONDS = AUTO_REFRESH_SECONDS - 60  # so autorefresh always refetches
 DISPLAY_TZ = os.environ.get("DISPLAY_TZ", "America/New_York")
 PLACE_NAME = os.environ.get("PLACE_NAME", "Yaphank, NY")
 
-# Two fixed views — no more dropdowns.
-VIEW_ZOOM = {
-    # 24h history is the longest window where Ecowitt reliably keeps 5-min
-    # data; beyond ~30h the cloud has already downsampled to 30-min and the
-    # zoom view stops being zoomed. 8h forecast keeps the chart ~75% history,
-    # ~25% future.
-    "label": "Last 24 h · 8 h forecast (5-min cadence)",
-    "cycle_type": "5min",
-    "resample": "5min",
-    "history_hours": 24,
-    "horizon_hours": 8,
-}
+# Single canonical view — weekly, hourly cadence.
 VIEW_WEEK = {
     "label": "Past 7 days · 72 h forecast (hourly cadence)",
     "cycle_type": "30min",
@@ -190,15 +179,6 @@ def _build_view(view: dict, log_conn, log_to_scoreboard: bool) -> dict:
     now = pd.Timestamp.now(tz="UTC").floor(resample)
     visible_steps = int(round(hours / step_hours))
     visible_history = history.tail(visible_steps)
-    since_unix = (
-        int(visible_history.index.min().timestamp()) if not visible_history.empty else None
-    )
-    past_toto: dict[str, pd.DataFrame] = {}
-    for m in METRICS:
-        col = m["col"]
-        pt = forecast_log.historical_predictions(log_conn, "toto", col, since_unix=since_unix)
-        if not pt.empty:
-            past_toto[col] = pt
 
     fig = combined_figure(
         history=visible_history,
@@ -206,7 +186,6 @@ def _build_view(view: dict, log_conn, log_to_scoreboard: bool) -> dict:
         nws_df=nws_future,
         metrics=METRICS,
         now=now,
-        past_toto=past_toto,
     )
     return {
         "fig": fig,
@@ -221,10 +200,7 @@ def refresh():
     realtime = fetch_realtime_snapshot()
     log_conn = forecast_log.connect()
 
-    # Weekly view is the canonical one logged to the scoreboard (hourly
-    # cadence keeps target_ts aligned with NWS hourly periods).
     week = _build_view(VIEW_WEEK, log_conn, log_to_scoreboard=True)
-    zoom = _build_view(VIEW_ZOOM, log_conn, log_to_scoreboard=False)
 
     # Hero uses the weekly history + the NWS period containing "now".
     nws_df_raw = week["nws_df_raw"]
@@ -250,7 +226,7 @@ def refresh():
     scoreboard = render_scoreboard(log_conn)
 
     persist.push_db_async()
-    return hero, comparison_md, zoom["fig"], week["fig"], scoreboard
+    return hero, comparison_md, week["fig"], scoreboard
 
 
 # --- scoreboard ----------------------------------------------------------
@@ -370,10 +346,7 @@ with gr.Blocks(title="Toto Weather Forecast", theme=gr.themes.Soft()) as demo:
 
     scoreboard_md = gr.Markdown()
 
-    gr.Markdown(f"### 🔍 Zoomed-in view — {VIEW_ZOOM['label']}")
-    zoom_plot = gr.Plot(label="Zoomed-in")
-
-    gr.Markdown(f"### 📅 Weekly view — {VIEW_WEEK['label']}")
+    gr.Markdown(f"### 📅 {VIEW_WEEK['label']}")
     week_plot = gr.Plot(label="Weekly")
 
     with gr.Accordion("How the scoreboard is calculated", open=False):
@@ -418,7 +391,7 @@ with gr.Blocks(title="Toto Weather Forecast", theme=gr.themes.Soft()) as demo:
             "Full spec: [`docs/toto-inference.md`](https://huggingface.co/spaces/bitsofchris/time-series-ai-weather-forecast/blob/main/docs/toto-inference.md)."
         )
 
-    outputs = [hero_md, comparison_md, zoom_plot, week_plot, scoreboard_md]
+    outputs = [hero_md, comparison_md, week_plot, scoreboard_md]
     demo.load(refresh, outputs=outputs)
 
 

src/weather_ui.py

@@ -187,8 +187,6 @@ def combined_figure(
     nws_df: pd.DataFrame | None,
     metrics: list[dict],
     now: pd.Timestamp | None = None,
-    past_toto: dict[str, pd.DataFrame] | None = None,
-    past_nws: dict[str, pd.DataFrame] | None = None,
 ) -> go.Figure:
     """Three stacked subplots sharing the x-axis."""
     fig = make_subplots(
@@ -214,18 +212,6 @@ def combined_figure(
             ),
             row=i, col=1,
         )
-        # Past Toto forecasts vs the same hours' actuals.
-        if past_toto and col in past_toto:
-            pt = past_toto[col]
-            fig.add_trace(
-                go.Scatter(
-                    x=pt.index, y=pt["p50"].values,
-                    name="🤖 Toto (past forecasts)", mode="lines",
-                    line=dict(color="rgba(31,119,180,0.55)", width=1.5),
-                    showlegend=showlegend, legendgroup="toto-past",
-                ),
-                row=i, col=1,
-            )
         if toto is not None:
             fig.add_trace(
                 go.Scatter(
@@ -268,7 +254,14 @@ def combined_figure(
     fig.update_layout(
         height=900,
         hovermode="x unified",
-        margin=dict(l=50, r=20, t=50, b=40),
+        margin=dict(l=50, r=20, t=50, b=60),
         legend=dict(orientation="h", yanchor="bottom", y=1.04, xanchor="right", x=1),
     )
+    # Explicit date + hour on the x-axis so the reader doesn't have to guess
+    # what day a tick refers to.
+    fig.update_xaxes(
+        tickformat="%b %-d\n%-I %p",
+        ticklabelmode="instant",
+        showgrid=True,
+    )
     return fig