Auto ARIMA Model Selection for Time Series Forecasting with pmdarima

This code is using the 'auto_arima' function from the 'pmdarima' package to automatically select the best ARIMA model for a given time series 'ts_log'.

Here's a breakdown of the parameters used:

• 'start_p=1, start_q=1': Starting values for the order of the AR and MA components of the model.
• 'test='adf'': The type of test to use for selecting the best ARIMA model. In this case, it's using the augmented Dickey-Fuller test.
• 'max_p=3, max_q=3': Maximum values for the order of the AR and MA components of the model.
• 'm=12': The number of periods in each seasonal cycle. In this case, it's assuming a monthly time series.
• 'start_P=0': Starting value for the order of the seasonal AR component of the model.
• 'seasonal=True': Whether or not to consider seasonal components in the model.
• 'd=None, D=1': The order of differencing to apply to the time series and the order of seasonal differencing.
• 'trace=True': Whether to print the progress of the model fit.
• 'error_action='ignore'': What to do if the model encounters an error. In this case, it's ignoring the error and continuing.
• 'suppress_warnings=True': Whether to suppress warnings from the model fit.
• 'stepwise=True': Whether or not to use a stepwise approach to search for the best model.