geocif 0.4.332__tar.gz → 0.4.334__tar.gz

{geocif-0.4.332/geocif.egg-info → geocif-0.4.334}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: geocif
-Version: 0.4.332
+Version: 0.4.334
 Summary: Models to visualize and forecast crop conditions and yields
 Author-email: Ritvik Sahajpal <ritvik@umd.edu>
 License: MIT

{geocif-0.4.332 → geocif-0.4.334}/geocif/__init__.py

@@ -2,6 +2,6 @@
 
 __author__ = """Ritvik Sahajpal"""
 __email__ = "ritvik@umd.edu"
-__version__ = "0.4.332"
+__version__ = "0.4.333"
 
 __all__ = ["ml", "cid", "viz", "agmet", "fdw_export", "dashboard"]

{geocif-0.4.332 → geocif-0.4.334}/geocif/analysis.py

@@ -189,6 +189,7 @@ class Geoanalysis:
         self.logger.info(f"Analyze {self.country} {self.crop}")
 
         self._plot_yield_with_ci(self.df_analysis)
+        self._plot_yield_with_ci_historical(self.df_analysis)
 
         df = self._clean_data()
         if df.empty:
@@ -700,6 +701,8 @@ class Geoanalysis:
         )
 
         median_col = "Median Yield (tn per ha) (2018-2022)"
+        if f"{median_col}_y" in df_ci.columns:
+            median_col = f"{median_col}_y"
         has_median = median_col in df_ci.columns and df_ci[median_col].notna().any()
 
         fig, ax = plt.subplots(figsize=(8, max(4, len(df_ci) * 0.4)))
@@ -736,6 +739,92 @@ class Geoanalysis:
         fig.savefig(self.dir_country_plots / fname, dpi=250)
         plt.close(fig)
 
+    def _plot_yield_with_ci_historical(self, df):
+        """Forest plot of predicted yield with CI and individual historical year yields."""
+        if "lower CI" not in df.columns or "upper CI" not in df.columns:
+            return
+        df_ci = df.dropna(subset=["lower CI", "upper CI"])
+        if df_ci.empty:
+            return
+
+        # Use latest stage per region
+        df_ci = (
+            df_ci.sort_values("Stage Name")
+            .groupby("Region")
+            .last()
+            .reset_index()
+            .sort_values(self.predicted)
+        )
+
+        # Load historical yields from statistics CSV
+        try:
+            if self.country == "pooled":
+                frames = []
+                for c in self.countries:
+                    f = utils.statistics_file_path(self.dir_out, self.method, c, self.crop)
+                    if f.exists():
+                        frames.append(pd.read_csv(f))
+                df_hist = pd.concat(frames, ignore_index=True)
+            else:
+                f = utils.statistics_file_path(self.dir_out, self.method, self.country, self.crop)
+                df_hist = pd.read_csv(f)
+            df_hist = df_hist[["Region", "Harvest Year", "Yield (tn per ha)"]].dropna()
+        except Exception:
+            return
+
+        # Last 5 years of available data
+        years = sorted(df_hist["Harvest Year"].unique())[-5:]
+        df_hist = df_hist[df_hist["Harvest Year"].isin(years)]
+
+        # Pivot: one column per year
+        df_pivot = df_hist.pivot_table(
+            index="Region", columns="Harvest Year", values="Yield (tn per ha)"
+        )
+
+        # Only keep regions that are in df_ci
+        regions = df_ci["Region"].values
+        df_pivot = df_pivot.reindex(regions)
+
+        fig, ax = plt.subplots(figsize=(8, max(4, len(df_ci) * 0.4)))
+        y_pos = np.arange(len(df_ci))
+
+        # CI error bars
+        xerr_low = df_ci[self.predicted].values - df_ci["lower CI"].values
+        xerr_high = df_ci["upper CI"].values - df_ci[self.predicted].values
+        ax.errorbar(
+            df_ci[self.predicted].values, y_pos,
+            xerr=[xerr_low, xerr_high],
+            fmt="o", color="steelblue", capsize=3, zorder=10,
+            label="Predicted \u00b1 CI",
+        )
+
+        # Historical year diamonds — each year gets a distinct color
+        cmap = plt.cm.RdYlGn
+        colors = [cmap(i / max(len(years) - 1, 1)) for i in range(len(years))]
+        for idx, yr in enumerate(years):
+            if yr in df_pivot.columns:
+                vals = df_pivot[yr].values
+                ax.scatter(
+                    vals, y_pos,
+                    marker="D", color=colors[idx], edgecolors="k",
+                    linewidths=0.3, s=25, zorder=5,
+                    label=str(int(yr)),
+                )
+
+        ax.set_yticks(y_pos)
+        ax.set_yticklabels(regions, fontsize=8)
+        ax.set_xlabel("Yield (tn per ha)")
+        ax.set_title(
+            f"Predicted Yield with CI & Historical Yields \u2014 {self.country} {self.crop}",
+            fontsize=10, fontweight="bold",
+        )
+        ax.legend(fontsize=7, loc="lower right")
+        plt.tight_layout()
+
+        fname = f"yield_ci_historical_{self.country}_{self.crop}.png"
+        fig.savefig(self.dir_country_plots / fname, dpi=250)
+        plt.close(fig)
+
     def _plot_national_yield(self, df_national_yield):
         from sklearn.metrics import (
             mean_squared_error,

{geocif-0.4.332 → geocif-0.4.334/geocif.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: geocif
-Version: 0.4.332
+Version: 0.4.334
 Summary: Models to visualize and forecast crop conditions and yields
 Author-email: Ritvik Sahajpal <ritvik@umd.edu>
 License: MIT

{geocif-0.4.332 → geocif-0.4.334}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "geocif"
-version = "0.4.332"
+version = "0.4.334"
 
 description = "Models to visualize and forecast crop conditions and yields"
 readme = "README.md"