Add models and complete forecasting service

README.md

@@ -79,21 +79,35 @@ make migrate
 
 ```json
 {
-  "product_name": "ginger_powder"
+  "product_SKU": "SKU001",
+  "days" : 30
 }
 ```
 
 ### Response
 
 ```json
 {
-  "product": "ginger_powder",
-  "current_stock": 120,
-  "forecasted_demand_next_30_days": 240,
-  "stock_shortfall": 120,
-  "daily_predictions": [
-    { "date": "2025-05-02", "predicted": 7.3 },
-    ...
+    "product_SKU": "SKU001",
+    "current_stock": 10000,
+    "average_forecasted_demand": 14608.27,
+    "maximum_forecast": 16425.71,
+    "minimum_forecast": 12764.46,
+    "stock_shortfall": 4608.27,
+    "daily_predictions": [
+        {
+            "date": "2025-05-03",
+            "predicted": 593.5,
+            "lower_bound": 530.5,
+            "upper_bound": 651.77
+        },
+        {
+            "date": "2025-05-04",
+            "predicted": 635.67,
+            "lower_bound": 575.94,
+            "upper_bound": 697.32
+        },
+        ...
   ]
 }
 ```
@@ -115,7 +129,7 @@ make migrate
 The service calls the external warehouse API:
 
 ```REST
-GET http://warehouse-service/api/stock/<product_name>/
+GET http://warehouse-service/api/stock/<product_SKU>/
 → Expected response: { "current_stock": 120 }
 ```
 
@@ -136,10 +150,4 @@ def prepare_features(df):
     return df
 ```
 
----
-
-## 🧼 Linting and Best Practices
-
-- Keep views thin — logic should stay in `forecast_runner.py`
-- Avoid public access to the `.joblib` model files
-- Use DRF (Django REST Framework) for extensible APIs
+---

forecast_system/forecaster/feature_pipelines/pipeline.py

@@ -0,0 +1,19 @@
+import pandas as pd
+
+def get_season(date):
+    month = date.month
+    if month in [12, 1, 2]:
+        return 'Winter'
+    elif month in [3, 4, 5]:
+        return 'Spring'
+    elif month in [6, 7, 8]:
+        return 'Summer'
+    else:
+        return 'Fall'
+
+def prepare_features(df: pd.DataFrame) -> pd.DataFrame:
+    df['is_weekend'] = df['ds'].dt.dayofweek.isin([5, 6]).astype(int)
+    df['season'] = df['ds'].apply(get_season)
+    season_dummies = pd.get_dummies(df['season'])
+    df = pd.concat([df, season_dummies], axis=1)
+    return df

forecast_system/forecaster/forecast_runner.py

@@ -1,28 +1,48 @@
-# import joblib
-# import importlib
-# import pandas as pd
-# from .utils.warehouse_api import get_current_stock
-
-# def forecast_for_product(product_name: str):
-#     # Load model and feature pipeline
-#     model = joblib.load(f"forecaster/models/{product_name}.joblib")
-#     features = importlib.import_module(f"forecaster.feature_pipelines.{product_name}")
-
-#     # Generate base DataFrame with dates
-#     df = pd.DataFrame({'ds': pd.date_range(start=pd.Timestamp.today(), periods=30)})
-
-#     df = features.prepare_features(df)
-#     forecast = model.predict(df)
-
-#     # Get current stock
-#     stock = get_current_stock(product_name)
-#     total_forecast = forecast['yhat'].sum()
-#     shortage = max(0, total_forecast - stock)
-
-#     return {
-#         "product": product_name,
-#         "current_stock": stock,
-#         "forecasted_demand_next_30_days": round(total_forecast, 2),
-#         "stock_shortfall": round(shortage, 2),
-#         "daily_predictions": forecast[['ds', 'yhat']].rename(columns={'ds': 'date', 'yhat': 'predicted'}).to_dict('records')
-#     }
+import joblib
+import importlib
+import pandas as pd
+import os
+from .utils.warehouse_api import get_current_stock
+
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+MODEL_DIR = os.path.join(BASE_DIR, 'models')
+
+
+def forecast_for_product(product_SKU: str, days: int = 30) -> dict:
+    model_path = os.path.join(MODEL_DIR, f"{product_SKU}.joblib")
+    if not os.path.isfile(model_path):
+        return {"error": f"Model for '{product_SKU}' not found."}
+
+    try:
+        model = joblib.load(model_path)
+        pipeline = importlib.import_module(f'forecaster.feature_pipelines.pipeline')
+
+        future_dates = pd.DataFrame({'ds': pd.date_range(start=pd.Timestamp.today(), periods=days)})
+        features = pipeline.prepare_features(future_dates.copy())
+        forecast = model.predict(features)
+
+        current_stock = get_current_stock(product_SKU)
+        total_forecast = round(forecast['yhat'].sum(), 2)
+
+        total_maximum = round(forecast['yhat_upper'].sum(), 2)
+        total_minimum = round(forecast['yhat_lower'].sum(), 2)
+        stock_shortfall = max(0, round(total_forecast - current_stock, 2))
+
+        # Convert dates to string format YYYY-MM-DD
+        forecast_copy = forecast.copy()
+        forecast_copy['ds'] = forecast_copy['ds'].dt.strftime('%Y-%m-%d')
+
+        return {
+            "product_SKU": product_SKU,
+            "current_stock": current_stock,
+            "average_forecasted_demand": total_forecast,
+            "maximum_forecast": total_maximum,
+            "minimum_forecast": total_minimum,
+            "stock_shortfall": stock_shortfall,
+            "daily_predictions": forecast_copy[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].rename(columns={
+            'ds': 'date', 'yhat': 'predicted', 'yhat_lower': 'lower_bound', 'yhat_upper': 'upper_bound'
+            }).round({'predicted': 2, 'lower_bound': 2, 'upper_bound': 2}).to_dict(orient='records')
+        }
+
+    except Exception as e:
+        return {"error": f"Failed to forecast for '{product_SKU}': {str(e)}"}

forecast_system/forecaster/urls.py

@@ -3,4 +3,5 @@
 
 urlpatterns = [
     path('', views.root_view),
+    path('forecast/', views.forecast_view),
 ]

forecast_system/forecaster/utils/warehouse_api.py

@@ -1,7 +1,28 @@
 # import requests
 
-# def get_current_stock(product_name: str) -> int:
-#     res = requests.get(f"http://warehouse-service/api/stock/{product_name}/")
-#     if res.status_code == 200:
-#         return res.json().get('current_stock', 0)
+# WAREHOUSE_API_BASE = "http://localhost:8001/api"
+
+# def get_current_stock(product_SKU: str) -> int:
+#     try:
+#         response = requests.get(f"{WAREHOUSE_API_BASE}/stock/{product_SKU}/")
+#         if response.status_code == 200:
+#             return response.json().get("current_stock", 0)
+#     except Exception:
+#         pass
 #     return 0
+
+def get_current_stock(product_SKU: str) -> int:
+    """Return hardcoded stock values for different products."""
+    stock_data = {
+        "SKU001": 10000,
+        "SKU002": 5000050,
+        "SKU003": 750000,
+        "SKU004": 500998,
+        "SKU005": 42000,
+        "SKU006": 585000,
+        "SKU007": 270000,
+        "SKU008": 45500,
+        "SKU009": 900000,
+        "SKU010": 4000000,
+    }
+    return stock_data.get(product_SKU, 0)

forecast_system/forecaster/views.py

@@ -7,3 +7,17 @@
 def root_view(request):
     return Response({"message": "Forecasting Service API"}, status=status.HTTP_200_OK)
 
+
+from .forecast_runner import forecast_for_product
+
+@api_view(['POST'])
+def forecast_view(request):
+    product_SKU = request.data.get("product_SKU")
+    days = int(request.data.get("days", 30))  # Default to 30 if not provided
+
+    if not product_SKU:
+        return Response({"error": "Missing product_SKU"}, status=400)
+
+    result = forecast_for_product(product_SKU, days=days)
+    return Response(result, status=200 if 'error' not in result else 500)
+