Scribd
Upload
19 views

exogenous variable hdl document

The document outlines a process for selecting and training forecasting models that incorporate exogenous variables, emphasizing the importance of model evaluation and forecasting accuracy. It discusses various methods for incorporating exogenous variables, including assuming constant values, forecasting them separately, conducting scenario analysis, and utilizing external data sources. Additionally, it provides Python code examples for implementing these methods in practice.

Uploaded by

Sunitha Menni
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
19 views

exogenous variable hdl document

The document outlines a process for selecting and training forecasting models that incorporate exogenous variables, emphasizing the importance of model evaluation and forecasting accuracy. It discusses various methods for incorporating exogenous variables, including assuming constant values, forecasting them separately, conducting scenario analysis, and utilizing external data sources. Additionally, it provides Python code examples for implementing these methods in practice.

Uploaded by

Sunitha Menni
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

1.

Model Selection: Choose appropriate forecasting models that can


incorporate exogenous variables. Time-series models like ARIMA, SARIMA, or
Prophet are commonly used for this purpose.
2. Model Training: Split the data into training and validation sets. Train the
forecasting model using the training data, optimizing model parameters as
necessary. Validate the model using the validation set to ensure it generalizes
well to unseen data.
3. Incorporating Exogenous Variables: Modify the chosen forecasting model
to incorporate the exogenous variables. This may involve adding them as
additional input features or integrating them into the model's structure,
depending on the specific model architecture.
4. Model Evaluation: Evaluate the performance of the model using
appropriate metrics such as Mean Absolute Error (MAE), Mean Squared Error
(MSE), or Root Mean Squared Error (RMSE). Compare the performance of the
model with and without exogenous variables to assess their impact on
forecasting accuracy.
5. Forecasting: Once the model is trained and validated, use it to make
forecasts of the quantity of interest based on future values of the exogenous
variables. Monitor forecast performance over time and refine the model as
necessary.

Elaboration on Incorporating Exogenous Variable-


1. Assume Constant Values: Assume constant values for the exogenous
variables in the forecast period. This assumes that the exogenous variables
will remain unchanged from their last observed values. While this approach is
simple, it may not capture potential changes or trends in the exogenous
variables over time.
2. Forecast Exogenous Variables Separately: If feasible, you can develop
separate forecasting models for the exogenous variables themselves. Once
you have forecasts for the exogenous variables, you can then use them as
inputs to your main forecasting model for the quantity of interest. This
approach allows you to incorporate the uncertainty in the exogenous variable
forecasts into your final forecast.
3. Scenario Analysis: Conduct scenario analysis by creating multiple forecasts
based on different assumptions about the future values of exogenous
variables. This approach can help you understand the range of possible
outcomes and assess the sensitivity of your forecast to changes in the
exogenous variables.
4. External Data Sources: Explore the possibility of obtaining future values of
exogenous variables from external sources such as economic forecasts,
industry reports, or expert opinions. While this may require additional effort
and resources, it can provide valuable insights into future trends and
developments that may impact your quantity forecast.

Situation 1: Assume Constant Values


Assuming last observed value of exogenous variable is used as constant
last_observed_exog_value = exog_var.iloc[-1]
forecast_constant_exog = results.forecast(steps=n_steps,
exog=last_observed_exog_value)
print(forecast_constant_exog)
Situation 2: Forecast Exogenous Variables Separately
python code
Assuming you have separate models for forecasting exogenous variables
exog_model = SARIMAX(exog_var, order=(1, 1, 1))
exog_results = exog_model.fit()
forecast_exog = exog_results.forecast(steps=n_steps)
forecast_with_exog_forecast = results.forecast(steps=n_steps, exog=forecast_exog)
print(forecast_with_exog_forecast)
Situation 3: Scenario Analysis
Python code
Perform scenario analysis by creating multiple forecasts with different
assumptions
forecast_scenario_1 = results.forecast(steps=n_steps, exog=scenario_1_exog)
forecast_scenario_2 = results.forecast(steps=n_steps, exog=scenario_2_exog)
print(forecast_scenario_1)
print(forecast_scenario_2)
Situation 4: External Data Sources
Python code
Assume future values of exogenous variable are obtained from an external
source
external_data = pd.read_csv('external_data.csv')
future_exog_values = external_data['future_exog_var']
forecast_external = results.forecast(steps=n_steps, exog=future_exog_values)
print(forecast_external)

You might also like