Update inference.py

inference.py

@@ -1,21 +1,18 @@
-import pandas as pd
-import pickle
-from datetime import datetime
+import sklearn  # Explicit import
+   import pandas as pd
+   import pickle
+   from datetime import datetime
 
-def predict_expense(input_date_str):
-    """Predicts the expense for the given date."""
+   class ExpenseForecaster:
+       def __init__(self, model_path="model/expense_forecaster_model.pkl"):
+           with open(model_path, "rb") as model_file:
+               self.model = pickle.load(model_file)
+           self.min_date = pd.to_datetime("2024-01-01")  # Reference date
 
-    # Load the model
-    with open("model/expense_forecaster_model.pkl", "rb") as model_file:
-        model = pickle.load(model_file)
+       def __call__(self, input_date_str):
+           input_date = pd.to_datetime(input_date_str)
+           numerical_date = (input_date - self.min_date) / pd.Timedelta(days=30)
+           prediction = self.model.predict(pd.DataFrame({"ds": [numerical_date]}))
+           return prediction[0].tolist()  # Return as a list
 
-    # Preprocess input (convert date to numerical feature)
-    input_date = pd.to_datetime(input_date_str)
-    # Use the minimum date from your training data (2024-01-01) as a reference point
-    min_date = pd.to_datetime("2024-01-01")
-    numerical_date = (input_date - min_date) / pd.Timedelta(days=30)
-
-    # Make prediction
-    prediction = model.predict(pd.DataFrame({"ds": [numerical_date]}))
-
-    return prediction[0]
+   model = ExpenseForecaster() # Instantiate the model