fix fallback strategy

[MartinoMensio]

"index" column does not exist. Reference the index from forecast dataframe

Trace:

KeyError                                  Traceback (most recent call last)
/tmp/ipykernel_68905/1675210467.py in ?()
      1 # TODO: check why values of past and future do not match (very different values)
      2 
      3 
----> 4 forecast = model.predict(historic_df=df_hist, prediction_df=weather_forecast_standardized, house_id=house_id)

~/code/energy-forecast/load-forecasting/demand_forecasting/forecast/model.py in ?(self, historic_df, prediction_df, house_id, known_ground_truth)
    183         id_full = f"{self.experiment_id_base}_{house_id}"
    184         pj = PredictionJobDataClass(id=id_full, **self.pj)
    185 
    186         # Make a forecast
--> 187         forecast = create_forecast_pipeline(
    188             pj,
    189             df,
    190             mlflow_tracking_uri=f"{utils.MODELS_PATH}/models/",  # {self.datetime_value}",

~/.cache/pypoetry/virtualenvs/load-forecasting-8OKFhNKY-py3.10/lib/python3.10/site-packages/openstef/pipeline/create_forecast.py in ?(pj, input_data, mlflow_tracking_uri)
     57     # Load most recent model for the given pid
     58     model, model_specs = MLflowSerializer(
     59         mlflow_tracking_uri=mlflow_tracking_uri
     60     ).load_model(experiment_name=str(prediction_model_pid))
---> 61     return create_forecast_pipeline_core(pj, input_data, model, model_specs)

~/.cache/pypoetry/virtualenvs/load-forecasting-8OKFhNKY-py3.10/lib/python3.10/site-packages/openstef/pipeline/create_forecast.py in ?(pj, input_data, model, model_specs)
    143             forecast_type="fallback",
    144             pid=pj["id"],
    145             fallback_strategy=fallback_strategy,
    146         )
--> 147         forecast = generate_fallback(data_with_features, input_data[["load"]])
    148 
    149     else:
    150         # Predict

~/.cache/pypoetry/virtualenvs/load-forecasting-8OKFhNKY-py3.10/lib/python3.10/site-packages/openstef/model/fallback.py in ?(forecast_input, load, fallback_strategy)
     64             forecast.reset_index()
     65             .merge(
     66                 highest_daily_loadprofile, left_on="time", right_on="time", how="outer"
     67             )
---> 68             .set_index("index")
     69         )
     70 
     71     # Rename so column is called forecast

~/.cache/pypoetry/virtualenvs/load-forecasting-8OKFhNKY-py3.10/lib/python3.10/site-packages/pandas/core/frame.py in ?(self, keys, drop, append, inplace, verify_integrity)
   6118                     if not found:
   6119                         missing.append(col)
   6120 
   6121         if missing:
-> 6122             raise KeyError(f"None of {missing} are in the columns")
   6123 
   6124         if inplace:
   6125             frame = self

KeyError: "None of ['index'] are in the columns"

[clara-de-smet]

@MartinoMensio We fixed an issue in the build that caused PR pipelines to fail, so I updated your branch to retry the building step. Sorry for the delay!
I did notice that the tests are not succeeding though, so you may want to take another look at the changes you made.