fix

routes/protected_prediction.py

@@ -18,7 +18,7 @@ def predict_auto(request: AutoPredictionRequest):
 
         series = df['amount']
 
-        result = auto_arima_forecast(series, forecast_periods=1)
+        result = auto_arima_forecast(series, forecast_periods=request.future_steps)
 
         return AutoPredictionResponse(
             rmse=result["rmse"],
@@ -47,14 +47,14 @@ def predict_manual(request: ManualPredictionRequest):
 
         p, d, q = request.arima_model
 
-        result = manual_arima_forecast(series, p=p, d=d, q=q, forecast_periods=1)
+        result = manual_arima_forecast(series, p=p, d=d, q=q, forecast_periods=request.future_steps)
 
         return ManualPredictionResponse(
             arima_order=tuple(result["arima_order"]),
             prediction=result["prediction"],
             lower=result["lower"],
             upper=result["upper"],
-            success=True
+            success=True,            
         )
 
     except ValueError as ve:

schema/prediction.py

@@ -8,6 +8,7 @@ class BasePredictionRequest(BaseModel):
     value_column: str='sold_qty'
     date_column: str='date'
     date_regroup: bool=False
+    future_steps: int=1
 
 
 class AutoPredictionRequest(BasePredictionRequest):

utils/statistic/auto_arima.py

@@ -6,7 +6,7 @@ import pandas as pd
 import numpy as np
 
 warnings.filterwarnings("ignore", category=FutureWarning)
-def auto_arima_forecast(series: pd.Series, train_ratio=0.8, forecast_periods: int = 1) -> dict:
+def auto_arima_forecast(series: pd.Series, train_ratio=0.7, forecast_periods: int = 1) -> dict:
     if series is None or series.empty:
         raise ValueError("Data tidak valid atau kosong.")