[QUESTION] Cannot replicate 2nd iteration results of my historical_forecasts() using predict()

[lubibi13]

I have this LinearRegression model.

model_tsmx_luz_temp = LinearRegressionModel(lags=[-1,-2,-3,-4,-5,-6,-7,-14,-21,-30,-60,-90,-120,-150,-180,-210,-240,-365],
output_chunk_length=365,
multi_models=False,
random_state=43)

-- temp_max_ser_luz ranges from 2021-01-01 to 2025-10-22

hfc_params_mult_luz = {
"series": temp_max_ser_luz,
"forecast_horizon": 365,
"verbose": True,
"retrain": True,
}

historical_fcasts_mult_luz = model_tsmx_luz_temp.historical_forecasts(last_points_only=False, stride=3, **hfc_params_mult_luz)

-- len(historical_fcasts_mult_luz) == 99

Using the predict method(), I was able to replicate the same results as in historical_fcasts_mult_luz[0]:

I reinstantiated model_tsmx_luz_temp again...

model_tsmx_luz_temp.fit(samp_temp_exa)
pred_tsmx_luz_temp = model_tsmx_luz_temp.predict(365)

THIS TIME, i tried to replicate historical_fcasts_mult_luz[1]:

I extended my training data (samp_temp_exa) by 3 days to simulate stride=3 in my historical_forecasts.

I reinstantiated model_tsmx_luz_temp again...

model_tsmx_luz_temp.fit(samp_temp_exa)
pred_tsmx_luz_temp = model_tsmx_luz_temp.predict(365)

As you can see, I cannot replicate historical_fcasts_mult_luz[1] which is the 2nd iteration result of my historical_forecasts. How did this happen??

[dennisbader]

Sorry, but this issue is not very easy to interpret. It would be great to improve this so we can help you :)

Can you please:

• replace the images by actual code
• provide a minimal reproducible example, that we can copy and run on our side to reproduce the issue
  • format code in code blocks (see here)
• state the complete issue in the beginning (e.g. what is going wrong as text)