From 4ef8fc5e9d54b2296306b5abf9bd6ef37de408df Mon Sep 17 00:00:00 2001
From: frasu <frasu@noreply.localhost>
Date: Thu, 10 Apr 2025 14:58:01 +0000
Subject: [PATCH] src/delta_barth/analysis/forecast.py aktualisiert

---
 src/delta_barth/analysis/forecast.py | 36 +++++++++++++++++++---------
 1 file changed, 25 insertions(+), 11 deletions(-)

diff --git a/src/delta_barth/analysis/forecast.py b/src/delta_barth/analysis/forecast.py
index 60b619c..5505613 100644
--- a/src/delta_barth/analysis/forecast.py
+++ b/src/delta_barth/analysis/forecast.py
@@ -231,6 +231,8 @@ def _process_sales(
         "early_stopping_rounds": [20, 50],
     }
 
+    # --- new: best_estimator
+    best_estimator: BestEstimatorXGBRegressor | None = None
     best_params: BestParametersXGBRegressor | None = None
     best_score_mae: float | None = float("inf")
     best_score_r2: float | None = None
@@ -238,20 +240,29 @@ def _process_sales(
     too_few_month_points: bool = True
     forecast: pd.DataFrame | None = None
 
-    # change sliding window to monthly basis
-    for start_year in range(current_year - 4, first_year - 1, -1):
+    # --- new: dates und forecast
+    #last_date = pd.to_datetime(monthly_sum.index[-1], format="%m.%Y")
+    last_date = pd.to_datetime(datetime.now().strftime("%m.%Y"), format="%m.%Y")
+    future_dates = pd.date_range(start=last_date + pd.DateOffset(months=1), periods=anzahl, freq="MS")
+    forecast = pd.DataFrame({"Datum": future_dates.strftime("%m.%Y")}).set_index("Datum")
+
+    dates = monthly_sum.index
+    for index, i in enumerate(range(len(dates)-36, -1, -12)):
+        current_date = dates[i]
+        split_date = dates[-anzahl]
+
         train = cast(
             pd.DataFrame,
-            monthly_sum[monthly_sum.index.year >= start_year].iloc[:-5].copy(),  # type: ignore
+            monthly_sum.loc[current_date:split_date].copy(),  # type: ignore
         )
         test = cast(
             pd.DataFrame,
-            monthly_sum[monthly_sum.index.year >= start_year].iloc[-5:].copy(),  # type: ignore
+            monthly_sum.loc[split_date:].copy(),  # type: ignore
         )
         X_train, X_test = train[features], test[features]
         y_train, y_test = train[target], test[target]
 
-        if len(train) >= (base_num_data_points_months + 10 * (current_year - 4 - start_year)):
+        if len(train) >= 30 + 10 * index:
             too_few_month_points = False
 
             rand = RandomizedSearchCV(
@@ -271,16 +282,19 @@ def _process_sales(
             if len(np.unique(y_pred)) != 1:
                 error = cast(float, mean_absolute_error(y_test, y_pred))
                 if error < best_score_mae:
+                    # --- new: store best_estimator
+                    best_estimator = cast(BestEstimatorXGBRegressor, rand.best_estimator_)
                     best_params = cast(BestParametersXGBRegressor, rand.best_params_)
                     best_score_mae = error
                     best_score_r2 = cast(float, r2_score(y_test, y_pred))
                     best_start_year = start_year
-                    # overwrite with pre-defined prognosis DF
-                    forecast = test.copy()
-                    forecast.loc[:, "vorhersage"] = y_pred
-    # remove
-    if forecast is not None:
-        forecast = forecast.drop(SALES_FEAT, axis=1).reset_index(drop=True)
+
+    # --- new: use best_estimator to calculate future values and store them in forecast
+    if best_estimator is not None:
+        X_future = pd.DataFrame({"jahr": future_dates.year, "monat": future_dates.month}, index=future_dates)
+        y_future = rand.best_estimator_.predict(X_future)
+        forecast["vorhersage"] = y_future  
+        
     best_score_mae = best_score_mae if not math.isinf(best_score_mae) else None
 
     if too_few_month_points: